"The Future of the Grid: Policy, Technology, and Market Change" by Casey Canfield [PWLConf 2018]
So I'm gonna be talking about the, future of the grid, policy. Technology, and market, changes. So. It's, an incredibly, exciting time to be talking about the changing, electricity, grid and so. As there's. A huge amount of change going on as we try to do this transition, from all of this 20th. Century, aging, infrastructure, to, try and meet our 21st. Century, demands, and, so. Obviously, this type of change is going to create conflict you know people are not gonna agree about which, way to go forward and so. Some places have. Like cities states counties they've, made hundred percent renewable, energy goals these. Are really trying to get at these sustainability, and climate change, issues. There's. Other places like, Puerto Rico, the Carolinas, just recently where, they're really focused, on resiliency. And just trying to figure out how can we keep the lights on in the face of all these increased, extreme weather events. And. Then. There's the utilities, the utilities. Are just trying to figure out how, they can keep operating how, their business, model can stay relevant and so, the utility death spiral, is something that comes up a lot in these conversations and. So the utility death, spiral, is basically the whole concept, that as, more, and more people are adopting, solar and storage and these types of distributed, technologies. They're. Gonna defect, from, the grid and so, then there's gonna be less people who are getting electricity, from their utility, and so, then the utility, doesn't have as many people to spread those costs across so, that's going to increase the cost of electricity coming, from the utility and more, people are gonna defect, and this is just going to go on and on until the utility, just you, know collapses, into a death spiral and. So. What. I want to talk to you about is, kind, of what's going on with the grid today on what, are driving, some of these changes and how are we going to navigate, this. So. A big question here is going to be, how. Like we have the what's going on today, I'm going to talk about one potential. Future a spoiler, alert it's gonna involve blockchain. But. Then the question that I'm really interested in is how. Do we navigate this, like what types of analysis, do we need for the regulators, and the policymakers and utilities, who, are the ones who are making these decisions and trying to figure out what should we invest in now and what's gonna be relevant, in the future and then, I want to leave you with a couple thoughts about how, you fit in and where you might be able to to, participate, in this.
So. Generally. Electricity. Is generated at. These big power plants, and. Gets. Shoved. On to the transmission, grid and so the transmission, grid is designed, to be able to transport, electricity over really, long distances. And so. Because, of that it's at a really high voltage, and so, it'll be at like a hundred, seven, hundred kilovolts, and. So then it goes through a transformer. And this steps down the voltage and so when you gets closer to your house it's actually, steps. Down to be part of the distribution grid and so the distribution, grid is at a lower voltage is, it like 10, kilovolts, and then. When it gets even closer to your house it. Actually gets stepped down to the 120 volts which is what we actually use, for a regular electricity. And. So. This. On this map of the transmission, grid you'll. See that there's some like gray and purple lines those. Are on the lower end of the voltage so that's around a hundred kilovolts. But. Then the red and the hot pink that's, on the higher end of the voltage and so that's they're going to be around three, hundred five, hundred kilovolts. And. So. This. System first, started coming online online, in the 1880s. So the very first, transmission, line was. Installed in 1889. In. Oregon. And so, was actually from. The. Willamette, Falls in Oregon City Oregon, to downtown, Portland, Oregon it was only 13, miles and so, by 1914. There were 55, transmission. Lines around the u.s., in. 1936. We passed the rural electrification act, and so that was to create these, low-interest. Federal, loans so that we could actually pay, for, distribution. Grids to be built in rural areas, and. Nowadays that, same laws actually being used for rural broadband, so that gives you a sense of our how. Our ideas about basic, needs have changed over time. And. So. The, entire grid is interconnected, but. It's also separated. Into three main parts and, so, there's the Western, interconnection, the eastern interconnection and, the Texas, interconnection, so, Texas, yes. Yeah. You don't mess with Texas and. So. These regions are managed, separately and within each region we, have several, different balancing. Authorities and so these balancing, authorities, are responsible for, balancing, the demand, and supply and the electrical, grid at every, single moment and so, right now we don't really have very much storage, on the grid which would you know make this a lot easier and. So. Operators. Have to be paying attention at every moment, forecasting. What electricity usage is going to be in the future to, make sure that the supply, and demand are perfectly. Balanced, in order, to avoid a blackout, scenario. But. Of course this, doesn't always work out the way we might hope blackouts. Are still a very frequent, problem, and so. Does, anyone remember the Northeast blackout of 2003. It. Was a major major, blackout, it, was also a really, really great example, of a cascading, failure and, so. Basically what happened is that it was a really hot August, day and.
He. Makes transmission. Lines sag and so. A transmission, line in Ohio touched a tree and it, caused the current to briefly increase, and this. Caused the protective, automatic, equipment to. Automatically. Shut off the line so. Then all of the electricity, that had been going down that transmission. Line needed, to find someplace else to go. And that caused other lines, to, get overloaded and then they tripped off and so. Normally this. Is a totally. Manageable. Problem, the. Operator. Pushes. A few buttons they, make a couple phone calls, everything's. Balanced again, however. In this particular situation. First. Energy who was, in charge of that first transmission, line it turned, out that is you might remember from earlier in the morning they, had a computer, that had a software, bug a race condition, and so. What what ended up happening was. That that computer, was frozen for about an hour before, anyone noticed and because. Of that no, one heard or saw any alarms. Suggesting. That everything. Was going horribly, horribly. Wrong and so. They didn't, intervene they didn't do anything and so. Ultimately this. Blackout, affected, 10,000,000, people in Ontario and, 45. Million people in, eight US states, so. All of New York City lost power took. Two days to two weeks to get the power back on Amtrak. Stopped running, cell, phone stopped working water. Systems, lost pressure. So. Obviously this was an incredibly, expensive, accident. Estimates. Put it at six billion dollars, and the. Reason I'm telling you this story is because, reliability. Is the most, important. Thing that a utility does and so, often, if you are talking to someone who works at a utility, and you have some crazy, new idea of how you're going to just revolutionize. The whole system, they just need to completely reorganize, no big deal, the. First thing that they're gonna tell you is I. Don't. Know I'm. Really worried that that's gonna impact reliability. Because. Utilities. Are not going to do anything that impacts, reliability. And they shouldn't having, reliable electricity. Is critical, for our economy, and for Public, Health. So. I've talked a little bit about how the, electricity, grids operated, give you a sense of that and so now I want to talk a little bit about the. Market regulatory. And policy parts, and so. As indicated. On this slide. States. Are generally, separated into regulated. And restructured. Markets, and so, in the white states like, Washington, and. Missouri, you'll. Find that we have regulated, gas and electricity markets, and so, generally what this means is that there's. A public. There's. A Public Utility Commission and, they're. Responsible for overseeing, a utility, who's a monopoly. That operates, in a specific region and so, if that utility. Wants, to, invest. In more infrastructure, if they, want to change the electricity, prices, they, have to go and get permission from, that regulator, and, so. In this context, you'll generally find vertically. Integrated, utilities, so, they own both the generation, equipment as well, as all the wires you'll. Also find smaller, utilities, - like municipal. Utilities, where they're, owned and operated by, the city government, also. Coops, so those are typically in rural areas, and they operate within not-for-profit. Business model and. So. The blue states are deregulated. Or restructured. And so, deregulated. Is kind, of a misnomer because, there's, actually still a lot of regulation. There's still a Public Utility Commission but. What this generally, means is that they're. Operated, with market, principles and. So. In, the dark blue states like Texas, and Oregon that's. Where they have, deregulated. Or restructured, electricity, markets then, the light blue states. Like. Florida, that's, where they've only deregulated. Their gas markets, and then, the stripy ones are where. We've deregulated. Both the gas and the electricity markets, and. So. In these, states what happens is that there's an energy, market to allocate, resources and. Different, companies, own the generation.
And The wires so, that no one has a, conflict, of interest and so. This is something this rewrite. Deregulated. And restructured, markets are still pretty controversial. So. For example there's lots of discussions, right now about how. Do a, hundred percent renewable, energy policy. Fit, into a market, structure um, the. Problem here is that you, can't just use a carbon, price because. These. Renewable. Energy policies, are not technology, neutral so. If you have a carbon price what's, more than likely to happen is that you'll increase natural gas generation, maybe, nuclear, generation, um but. You won't necessarily drive. Solar, and wind adoption. The way that the law is really intending, so. Then the question is okay but carbon, price isn't going to work what, are we supposed to do should we have separate. Markets for renewables, should, weari regulate, I mean this is a serious conversation that. People are having there's, lots of uncertainty. About where we're gonna move forward. So. Just, to give you one example on. The policy, side I'm. Gonna talk about how, net excess, generation so, basically if you have solar panels on your roof and you produce more electricity than you use that's, excess, generation, and the way that that gets compensated. Differs. Depending on what state you live in and so, in the. Dark blue states like California. The. Excess, generation is. Compensated. At the retail rate so, basically this means that, if. You are generating. More electricity, than you use your. Electricity, meters actually going to run backwards. And. So that's awesome if you, really want to encourage people to adopt solar it makes the economics, awesome. But. The negative side is that it makes the utilities, job a lot harder because. The entire electrical, grid is designed, so, that utilities, put power on and customers. Take power off it's, a one-way flow and so. With solar power, we're completely, changing, that whole paradigm so. Now customers. Can, inject power into the grid and, at. The same time the utility, is still, responsible for reliability, they're, still responsible, for making sure that everything, is perfectly, balanced, that whole time and there, is still a cost associated, with that and so. At the like middle blue states, so. For example Maine what, they do is they and say at the retail rate at first for, some of the generation, and then they switch to a lower rate, at. The light blue like, Missouri they. Only will credit it at a, lower rate typically, it's the avoided cost rate or the, wholesale, rate and so this is basically what.
The Utility, would have paid for that generation, and the, idea here is that it's. Actually accounting. For a little bit of the cost that the utility, is incurring because. There's still the backup generation, for, that home, so. Even if you, are putting electricity, into the grid at any moment your. Solar panels for example might stop working and you would need electricity, and the utility, is there to make sure that you get it and then. In the white states like Texas, they, are sorry in the stripe ones which Oklahoma. Is the one example here, they, actually have no obligation. To compensate excess. Generation and, then, in the white states like Texas, they, don't have any statewide, rules. Obligating. Them to compensate. Or not compensate, generation. And. So. The main point I wanted to make here is that, when. You're trying to figure out the economics, of solar it's not necessarily, just about how, sunny your roof is there's. Going to be different, market, and policy. Conditions. That affect the economics, of it and this varies, depending, on what state you're in. And. So. I've talked a little bit about how, the electricity. Is operated, today and how there's this really complex, interplay. Between. Technology. The market, is the regulatory, environment and policy. However. I also mentioned, that everything's, changing, and so. There's three trends, that are really shifting, the industry, so, that's decentralization. Electrification. And. Digitalization. So I'm going to go into a little bit more detail about each of these. So. Decentralization. The whole idea is that we're shifting, from this really centralized. System, where you have these big power plants, they're distributing, all the electricity, to a decentralized. System where. Anyone. Can go and buy some solar panels, and be injecting, power into the grid and. So. The main technologies, that are involved here are distributed, generation so, typically solar panels, distribute. Its storage, but, then also using. Less electricity matters. Here so negawatts. Is typically. The term a phrase so, energy, efficiency, and demand response and, so. So. And. So these different technologies, create, problems, as well as opportunities, so. For example with solar panels one. Of the complaints, that utilities. Often have is that, there's this thing called the duct curve and so. It's called a duct curve because, people think it looks like a duck so. If you see the duck. Yeah. It looks like a duck, and. So. The idea here is that when. You have this sharp, drop, in. Electricity. Usage, that's, really hard for utilities, to deal with these, big power plants, like coal and nuclear they. They can't change the electricity, output that quickly and so that's why there's been a big resurgence and why it's been really valuable to have big resurgence in natural gas because. That has a much better ramping, capability.
And. So that's why people are getting really excited, about storage, and so, storage, has. The potential to address the, duck curve and so, storage, by. Being. Able to absorb some of that electricity, can flatten the, peaks and valleys and, so, what this means is that utilities. Are super happy because it actually makes the electricity, much, more predictable, or that electricity demand, much more predictable, and utilities. Like things that are predictable, it makes their job really easy. But. Then there's also energy, efficiencies, the idea here is to reduce the overall demand, and so this reduced the number of power plants, that utilities, need to build, and. Then there's demand response and so this is typically a very targeted, effort. And so, for. Example when during, a peak. Demand crisis, so for, example it's a hot summer day everyone's. Got their air conditioning running and. The utilities, having, trouble supplying, all of that electricity, they'll. Use Adam Anderson, program, to. Get people to reduce their electricity, usage, so that they can keep the supply and demand perfectly. Balanced. And. So. In each of these cases customers. Are making decisions about. Whether. Or not to implement these technologies and, utilities, may or may not be involved in that decision and so, it's become really clear utilities, that they need to be a lot more involved in customer. Decision-making and being able to coordinate across, customers. So. The next big trend is electrification. And, so, in this graph on the Left this is talking about how the, cost of lithium-ion batteries. Has been dropping and is projected to continue to drop and, so. This is making storage, a lot more of a realistic, option and, so. In the graph on the right you can see that electric, vehicles are projected, to be an increasing. Percentage of, new car sales and so, by 2040. It's almost 40 percent of new car sales and so, this creates a lot of really, exciting, opportunities, and so, for example instead of having homeowners. Buy, a separate, storage, a, stationary. Battery storage for their house maybe. They can just use their car that they already have you know in this future where everyone has electric cars. You. Know maybe we. Can provide, grid, reliability services. Buying. Or stopping the charging, of electric vehicles. Based. On what's happening on the grid and what's going to provide the most benefit, you. Know maybe if we really think far, into the future if we, have autonomous, vehicles, maybe, while you're at work you're, autonomous, electric, vehicle, can, go drive to a part of the grid that needs some support and, provide. Services, there and make money while you're at work you. Know sky's limit there's. All kinds of crazy things. But. The idea here really is that electrification. Is, really, changing the boundaries, of what matters for grid operations, it's. So. Utilities. Worry about, people using less electricity because, it affects, their bottom line but, also it. Affects their, degrees. Of freedom for being able to address grid, reliability so. Electrification. Increases. Those degrees of freedom. And. The. Last big trend is digitalization. And so. The idea here is, that we, have all these smart sensors, and. Automatic. Control systems, and all. This date big data being collected, that. We can use to optimize the, system and. So. For. Example there's some utilities, who, when there's a blackout, they have no idea how big it is they have no visibility, into their system and that's why you have to call them on the phone and tell them hey, I don't have electricity because they don't know and so. There's been a big push to have advanced, metering, infrastructure, or am I on smart. Meters on this it's been a big part of pushing. The smart grid in the last 10 years and. So what this does is this gives utilities, a lot, more visibility, into, the system so that they know how, much electricity individual.
Houses Are using and so. This also means the utilities, are collecting. Huge, amounts, of data about how, much, electricity homes, are using and most, utilities, don't, even have the capacity to deal with any of that data like they don't have a data scientist, you. Know they're barely able to just store all that data, so. There's a lot of opportunity. To be able to better leverage that that's not happening, right now this. Also includes, on technology. Like smart, inverters, so, smart, inverters, are. Basically. A piece of electronics. That interfaces. Between a solar panel and the grid so. We can imagine that some, point we'll have these smart inverters, where. They. Can actually sense, the grid conditions, and when, they're injecting, power they, can change, what they're injecting, based on the conditions of the grid to increase, grid reliability, and so, that's actually how we get to some of these ideas about, having, a self-healing, grid. And. So. Of course all. Of this very exciting, but, what people worry about is the, cybersecurity implications. And so, there's already cybersecurity. Risks, at utilities. And this is only going to increase them and so, there's a lot of work being done for example to, increase information sharing, across utilities, so. That they're, better, able to identify attacks. Sooner, in, some. Ways you might think that oh this, might be an argument for, why we should, have more distributed, energy generation because. If the system's more resilient, then, that means. We. Won't be, the. Impact of a cyber attack won't be as bad and so then it won't matter of, course that assumes that, the. Smart inverter which is a critical piece of a distributed energy, system is not, the actual attack vector, so, there's still a lot of things to get figured out. So. I've talked a little bit about the grid today and hope I've sort. Of explained, how it's a pretty complex, system, right now and it's, only gonna get more and more complicated, as we move into the future and so. Now I want to talk about just one. Example future. That we could be headed towards. So. Everyone close your eyes close. Your eyes and. So. Let's. Imagine a, future, where, you just bought a house in a new up-and-coming, neighborhood.
The. House already, has solar panels and a built in home, energy management system. Which. You're well aware of because, it made it one of the more desirable, houses, on the market and you definitely paid, a premium for it. You. Pull your electric, car into, the garage and, it starts wirelessly. Charging, based, on your settings to minimize cost you. Start plugging in all of your Internet of Things devices, of course. They won't talk to the home energy management system, right away you, didn't die and go to heaven this is the real world. You. Spend a couple hours debugging. It and now the management, system can control your fridge your, air conditioner, your dryer a couple, other appliances. Based, on the real time price signals coming from the grid you. Check your mail and you have a welcome to the neighborhood block, chain letter that's. Right your cool. You have a neighborhood, micro-grid, with a peer-to-peer market. So. You can open your eyes so. Let's, explore that a little bit like what does that mean to, have a neighborhood, blockchain. And. So. Decentralization. Electrification. And, digitalization. Aren't, just changing, the technology. Of the grid it's, also forcing, people to completely, rethink how, the market, and policy, environment, is and. So. One of the more interesting departures. Is these, block chain based micro, grids or more, broadly we can think of them as local energy markets and so. The idea here is that there are consumers, who are homes without any generation, and then, there's prosumers. Who, are consumers, who are producing, electricity, and. There's. Some sort of market mechanism, that manages, the flow of information, about energy transactions. As well as the financial, transactions, and so. There's. A lot of interest in the potential of blockchain, to be that market mechanism. The. Idea is that this, could reduce the administrative cost, of transacting, at this scale assuming. We get a couple minor. Issues, like, the computational, cost of blockchain, figured, out but. That hasn't stopped people from using it and. So. Just in case you're like the one person who. Doesn't know blockchain is I'm gonna do a quick quick review so. The, idea is that there's, a participant, who, is, in the network and they want to request a transaction, of some sort and so, that request is broadcast, to the other computers, in the network and then, each computer, has two independent, independently. Validate. Or authenticate, that. The request is valid then, once. It gets verified, you. Can produce a new block of data that. Data. Is added that block is added to the blockchain permanently. And you've. Done it. And. So. It's, anyone here from Brooklyn. Have. You heard have you heard if the brooklyn micro grid. Okay. It's not totally, surprising I guess so. I don't have any like special, insider, information about the Brooklyn micro, grid, my understanding is that's a pretty small project, it's. Run out of the lo3 energy. They're, a start-up and. So my understanding is that there's only a couple people who are transacting. On it they have these big ambitions, to do real-time pricing and they haven't quite gotten there yet but. It's kind of exciting that, you know people in the US you, know not not just in Europe are. Actually, testing. Using. Blockchain. For, a local, energy market, on a micro, grid here. And. So. Generally. The idea is, that you, have to separate, the virtual, and physical layers, and so, the virtual layer, is the blockchain part, so, all of the market. And pricing, information is, added to the blockchain on, one block at a time and, transmitted. To all of the other homes on the network. However. In order for this to work you need to also be able to, transfer. The electricity. You cannot transfer electricity, with blockchain. So. For, this part just. To be clear, and. So, for this part they actually have to take advantage of the distribution, network and so in this case they're working with their local utility, ComEd and. ComEd. Is. Facilitating. This transaction because they're the ones who actually own, the wires and so, what this means is that if you're trying.
To Buy renewable. Energy from the, neighborhood, blockchain. In the Brooklyn micro grid you're. Not guaranteed. To. Get electrons, that came from solar panels that's, not how the grid works you can't control, where the electrons, go so. What you're really paying for is for, someone to generate, electricity from a solar panel and put, that onto the grid and then. You take electricity, at that same time and. So as more, and more homes get solar power in your micro grid the, probability. Of the, electrons, you personally, receive being from solar power increases. So, over time you'll get there but there's no guarantee. And. So. How exactly does, this transaction, work so. The idea is that you have consumers. Who are the indicated. By a see those, are the people who need to buy electricity. And then you have the prosumers, who are, designated. By a P and, they're the ones who are producing electricity and so. Typically. You'd have an automatic system that does this but basically each, consumer, needs to put. In an order for how, much electricity they, want to buy in the next fifteen minute interval, and then what's the maximum, price that they're willing to pay then. Each prosumer. Needs, to put in an order saying how much electricity they're, willing to sell in the next fifteen minute interval and what's, the minimum, price that they're willing to accept. Then. Whoever. Said they were willing to pay the most for the electricity they, get their electricity needs allocated, first and so. Once. They. Get their electricity needs, and you go the next person, the next person the next person and then the last person, who gets, electricity. Allocated, to them based on how much is available that's. The market clearing price and so. That's the price that everyone pays and, so. This. Is actually really, really similar, to how energy, markets work at a regional, scale in general. And. So. One, of the main barriers here, is that there's, actually no legal. Or regulatory structure. For, this sort of hyperlocal transaction. And so this is one of the major things that the Brooklyn micro grid is really trying to figure out like what should the rules be like, what license, do they need like. Whose permission do they need they're, really trying to figure out some of these questions and so it's it's kind of exciting to see, where this goes so I encourage you to follow, the Brooklyn micro grid see, where they go.
But. Of course not, everything, is super. Shiny and wonderful there's. Definitely a lot of people who have some serious concerns about using blockchain, for. Energy transactions. There's. Concerns about the, electricity, usage the computational. Costs that maybe this is going to wipe out the. Cost savings you get from, the reduced transaction, costs, maybe, this is going to be too complicated for your average consumer. There's. Concerns about security, and privacy. You, know how, much information do you want your neighbors to have about how much electricity you're using what. If a company had that information does that change things so. There's still a lot of open questions and technical, questions, that, I'm very confident, you know people will sort out over time and so, a bigger question that's. Much harder to answer is is, this, the best idea, like, is this where we should be, going with our electricity, system. And. So. I want to talk a little bit about that about these. Are some of my personal research. Interests, about how do we figure out today, how. To set ourselves up for, the best future. And. So. We've. Really only discussed, one potential, future in one, regulatory. And policy environment. And as, you may recall the. Rules are really different depending on what state you live in and. So. Depending, on where you are the, rules about how the electricity, grids operated, who can participate and, how, they participate, or going to be really different and so, there's not going to be a one-size-fits-all, solution, you. Know not everyone, is going to think a blockchain, based, micro, grid is right for me and so. For example like thinking about some of these equity, questions. You. Know there might be people who can't afford the capital, costs, of these home energy management systems, that, they then also need to program that's a lot of work setting. Up all your settings in what you want it to do they. Might want to just know what their electricity bill is going to be and not have to worry about any fluctuation. From markets. And. So we need to make sure that low, and middle income households, in particular are able to engage in these futures and that we're not just setting it up for the elite. And. So. One, proposal, for how to think about these. Futures, is from. The National Labs through. The Department of Energy and so. This, walk, jog, run model is set up to try and reduce risk, because. As I mentioned before the, most important, thing in the electricity system is. Make sure that it is super, reliable, and so, in, that, sort of system it's really hard, to deal with sudden. Change this, is not a model that's, set up to do disruptive. Change it's, about having a slow evolution, so. For example right now we're at stage one so, we have really low distributed. Energy. Adoption. We, have aging infrastructure, we're. Still just trying to get reliability. And resilience, and efficiency, in place, but. Eventually, we'll get to stage two and that, point we're jogging and so. The, idea here is that we're going to have a little bit more distributed. Energy resources, adoption. And. At that point really gonna have to rethink how. The distribution. Grid works entirely, so. Right now the distribution, grid is really focused, on just delivering, power it's one job deliver, power, but.
We Really are going to need to think of it as a platform as, something, that's providing, Grid services, and ensuring. That, electricity, is reliable. And dealing, with all of these injections. Of power at. Different points in the network and. Then. Eventually even later we're, going to get to stage 3 where, we're running and the, idea here is that we'll have really high distributed. Energy adoption. And. That's when we're gonna see, really different, utility. Business models, we're gonna see more local, energy markets it's going to be a really different world and the, idea here is that this, type of transition, gives. Us time to get, a new status, quo at each step and this, is what's going to make sure that we maintain reliability. In the system. So. I want to take a step back and sort of talk about how I approach these types of things and so, my academic, training is really focused, on, being. Able to quantify the human, part of complex, problems and so, I assume, that for any complex, problem, there's, going to be human, organizational. And technical components. To both the problem, as well as the solution, and. So. The main way that I address. This is using decision science and any decision science, project. Is going to really, be focused on one or more of these three questions. So how should people make decisions you do normative, analysis, how, do people, make decisions with. Descriptive, research and then, how can we help people make better descriptions. And this, is really looking at prescriptive. Interventions. And, things. Like randomized controlled, trials, which I'm going to talk about a lot right now and. So. As. We go down this path what. I'm. Arguing is, that we. Don't do a very good job of making sure that we have an evidence-based. Strategy. We, don't do a very good job of making sure that we're building a body of evidence to. Make decisions, based on and. So. I would argue that the, best way to build, that body of evidence is with experiments. And I. Don't mean experiments. As in try it and see if it works. I mean, randomized. Control, trials, and we should run them the way we were on medical trials. So, this is kind of getting at one of those subtle differences, between engineering, and science, this, isn't necessarily about testing. It to see if it works it's, about building a body of evidence and so. I'm going to talk about two, papers, to really drive this point home on the importance, of designing scientifically. Valid, experiments, at this intersection of. Technology. And policy and, markets and so, I'm going to be really focusing on consumer facing experiments. Because, whenever you involve people things. Are more complicated and therefore more interesting. So. It take a step back, why, are randomized, control trials the gold standard so. The whole idea here, is that you, have a population, of interest and, then, you randomly assign people to either the intervention group which might be receiving.
A Programmable, thermostat or the. Control group which would be receiving a regular, old dumb thermostat. And. So the idea is that you collect data and you measure outcomes, and at that point you can test your hypothesis, that, people who got the programmable. Thermostats, are going, to use less electricity and, you. Can make a causal, statement about. Their, programmable, thermostats, because. You randomly, assign people into, those two groups so, the whole idea is that. Any, other, reason. For people to use electricity, it's going to be randomly, distributed, between these two groups and so, any change, is going to just, be because of the programmable, thermostats, the. Other key piece here is the control group and so. It's. Possible that, because, of the economy, or something like that people, are going to reduce their electricity usage, anyway and so, you want to be able to account for that when you're trying to estimate the effect of the programmable, thermostats. And. So. You, know maybe that sounds easy but it's, not, it's. Actually really complicated to, run, a really good, randomized. Control trial and so. Alex, Davis and colleagues at Carnegie Mellon University. Examined. 32, publicly, available pilots. That estimated, the effectiveness, of three interventions, to reduce residential. Electricity, usage, and these, included, in-home displays, dynamic. Pricing and, automation. Like programmable. Thermostats, and so this was actually the, research group that I was in when I was doing, my PhD at, Carnegie Mellon. And. So. There's. Six types of bias that, we know about for medical trials, then we need to watch out for in randomized, controlled trials, and so, basically. It. Gets down to. A. Really. Poor randomization. And so, if. You have. People actively, volunteering to participate in a particular experiment, it's, biased if they're. Assigned to an intervention, or control group that's not completely random it's, biased even, some, things that might seem sort, of random like, assigning, people based on their birthday, not, random, there might be something systematically. Different about people who are born in June, rather, than November, and. So. If people know that the other conditions, and what the other conditions in the experiment, are it's. Also going to be biased and so. These, are the types of issues that we really need to be thinking about when we're designing, randomized, control trials and, so. Davis. Alex, Davis and his colleagues. Looked, at these 32 publicly. Available electricity. Pilots, and they found that most, of them were biased or they. Weren't documented, well enough to, even be able to tell if they were biased which. You know makes everyone pretty suspicious that, they were biased and so, one of the worst ones was. Volunteer, selection, and, so. In this case, 27. Out of 32 of, these studies used an opt-in, design rather than an opt-out design, and so, what this means is that they, were only doing, these experiments, with people, who volunteered, and wanted to participate and those, are the people who are most likely to benefit, from the intervention and so what's gonna happen is you're going to inflate, your results. And. So. For example what they found was that after, is that they, use, some statistical, methods to estimate the. Impact of the biased. Study design and they, found that for, peak reducing, peak electricity, use so that's the maximum amount of electricity, that. A home uses, the. Results. Were roughly halved so, they were still significant, but, they were just smaller and the may the one that works the best was, dynamic. Pricing with automation and so that people. Reduce their electricity usage, by fourteen percent as opposed, to the original, study which, estimated, it at twenty five percent. Also. Looking at overall electricity. Use they found that most, of the interventions, that were tested the original, studies found really small effects. But. When they added the Corrections, they weren't statistically. Significant, there was no impact, I'm the, only one that still remains standing was in home displays, and so, on they, were significant, but they had a really small effect so, it dropped to three percent from, five percent and this, is really important, if we're thinking about how, to do cost-benefit. Analysis. Because. If we are overestimating the benefit, from these types of programs then. It's. Going to be really hard, to.
Figure Out how much cost. To actually allocate to, running them in the real world. Also, company, politics, can get the way when. I was doing my PhD we, tried, to work with a utility, to run a randomized, control trial looking at in-home displays. And. We wanted to do an opt-out, design, because it's the gold standard and. We, explained that oh this would mean that we take a thousand, customers we'd, send them in-home displays, and we'd explain that it's an experiment, and. They can opt out if they want but. The utility was terrified, to implement this they were really worried that they'd get terrible, press, because. People thought that they were wasting, money because, they were sending in-home, displays, to, people who didn't want or need them. And. So what they ended up telling us was that you know we, how. About we tell you what we want to do and then you tell us how to do it as scientifically, as possible. Obviously. That's not a way to do good science, so. We ended up not being able to work with them and we never ran that study and, so. There's a lot of challenges to, kind of getting utilities, on board with this notion that randomized, control trials, are the right way to go. And. So, one. Of the challenges is even if you lay it out in this nice way where you explain that there's technology pilots. And then eventually you involve you evolve to demonstrations, where you engage consumers, and I'll, just point out that I'm not using the word pilots, in this way I'm using pilots, for everything, and I'm not gonna stop doing that because I showed you this slide. So. One of one of the challenges, is that you, know sometimes the point. Has not doing a pilot has nothing, to do with science in. Some cases utilities. Are just trying to keep their Public Utility, Commission happy, you know they just want to look busy and avoid, changing, the status quo, they, have no intention, of learning anything with these pilots, in, other cases there's, just a lot of confusion about who's, responsible. For deciding, if it's good science, or not like, is it the public, utility, Commission's, job like, these are a bunch of lawyers they don't they don't necessarily know, what's. Good science, and what's not and so, there's a lot of challenges associated with making sure that utilities, are actually, engaging with the right kinds of experts, when they're trying to make these decisions and build, this body of evidence. So. One other paper that I do want to mention is. Called. Solarize your community this, is actually, a public facing report which i think is awesome and everyone should, try and condense like six. Academic papers, that they've written into, a public, facing report, that they can actually share with people, and. So, this is by Ken Gillingham, at Yale, and Brian Bollinger, at Duke University, and. So what they were doing was that they, saw. That there were these solarized, campaigns, and so Solarize is basically when a city or town all gets, together. And. Gets. Lots of people to adopt solar, at the same time so that everyone gets a price discount. And. So they were able to take advantage of this rollout happening, in Connecticut, you can see there's a big, cluster, in Connecticut. To. Actually run randomized, control, trials, and learn something about what's the best way to design, a, solarized, campaign. And. So. They found that, having. A solarized campaign. Increased adoptions. So the Solarize campaign, is the blue line and the. Control. And the non solarized, adoptions, are the red line and the gray box is showing how long the solarized campaign, lasted, and so, you can see that, there's a huge increase in the number of solar adoptions, that happened, at that time and, that there was also a really sharp decrease, in the, price that people were paying for those installations, and that, that price decrease, only lasted, as long as the Solarize campaign lasted, so there's a lot of incentive, for people to participate. And. So. They've tested, a bunch of different variations, to try and figure out you. Know what is it about the solarized model that really makes it work you know what's, the core, of it that matters and so, ultimately what they found was that the choice model, was, the best approach and, so. It had a competitive application. So cities and towns are really motivated they had resources, available to devote to it it lasted.
For Twenty weeks which, was enough time to kind of roll it out when, they limited it to ten or twelve weeks they actually increased, administrative costs. Because. It was just so much harder to get the word out that quickly, it. Had a tiered, pricing structure, so. As more and more people adopted. The group, discount, that everyone got got. Got increased, got higher and higher and so people were really motivated to get their neighbors and other people to adopt solar so, that they'd get all get a better price, but. The key was that they had two to three installers, so, introducing. This small amount of competition was. Really valuable and, they, actually had the most adoptions. And the lowest, prices across, all of the models that they tested, and also. That price decrease, lasted. The longest after, the, solarized, campaign, ended and so, this little adding, just this little tweak of adding. Some competition, among the installers, had a huge impact and they never would have known that unless they had run, all these randomized, control trials, and. So. This. Has really gotten me thinking about you know how can we do a better job of rolling out this information so we run these randomized, control, trials, and then how do we make sure people actually get that information. And so, one, thing I found out about was this whole concept called implementation. Science and. So it's something that's really mostly, done in public health and it's basically this whole body of research they, have a journal, they have a conference, to. Try and figure out after. What happens after, you do the randomized control trial how, do you once you figure out the best way to do something how do you get everyone to do that how do you get all the doctors and all the hospitals, to actually do that and so, I think that there's an analogy, here to energy that we need to think about how to do this for the electorate's the electricity, system. And. So. I'm just you know spitballing, some ideas you know I think that you, know we can do research on regulator. And utility decision-making it's something I'm interested in, we. Can aggregate the results of experiments, and pilots and demonstrations, and actually, actively, build this body of evidence. We. Maybe we should be training utilities. And regulators and policy makers on what it means to do randomized, control trials and why they're valuable. Wouldn't. It be awesome if if utilities, had a chief evaluation. Officer who's responsible, for making sure that this stuff happened, that'd, be awesome, and, so. Last. I just want to leave you with a couple, thoughts about you know where you fit in. So. We've talked about how the grids interconnected. How, we have this complex. Interplay. Between the regulatory market, and Pauwels the environment, how. There's these trends in decentralization. Electrification, and digitalization. That are driving a lot of trains it's. Possible. That blockchain, base micro grids are our future. People. Are advocating, for this walk jog, run model to have this slow evolution in, electricity, system but. If we're, doing that we need to make sure that we're running scientifically. Valid, experiments. And especially. For customer-facing programs, even, though that's very politically. Fraught and utilities. Are scared. Because, most of the electricity pilots, that are happening are biased, and ironically, we're actually going to waste more money in the long term running, these biased, experiments. Than. If we just did it right from the beginning and so. The the, real trick here is to use randomized, control trials, whenever. We're running rolling, out programs to actually optimize the program design and.
So. What. I want to leave you with is that you know there's opportunities, for you to engage with some of this you, know you can participate like. If you own your house like you can think about getting, solar if. You can go to energy, sage calm, this. Is a really great resource for figuring, out if solar makes sense for you you know it's your electricity, system you can participate. Also. Advocate, you, know a lot of these discussion, of these decisions, are being made by policy, makers and elected officials so. Make sure that they know what kind of electricity system that you want you. Know make sure that you vote it's, always a good idea. If. You are in the electricity space or you're thinking about getting into it you, know run experiments, it's really really important to make sure that we're building this body of evidence and sharing, it and then. Lastly I just want to emphasize the importance, of increasing accessibility, there's. A lot of opportunity. To make sure that this transition, is accessible, as, particularly. For, low and moderate income households, where. They might not have the, financial, or, just, time resources. To, engage with, all of this change and so, you know technology's, not going to solve all of these problems but. It's definitely going to be part of the solution, so thank, you.