 been on the right hand side of the screen. If you care to please give us your thoughts on the topics that you're most interested in seeing at future webinar, we're asking that you select your top three and we'll take that as input when we're crafting future topics and panelists, speakers for those sessions. Thank you again for joining us today. We're really excited to have everybody here with us and wanted to note coming into the meeting that, as you entered the meeting, the webcast, everybody is auto muted. You can ask questions throughout either through the chat function and you'll see on this little slide illustration, it'll show you that at the bottom of your screen, you'll see a little like cartoon bubble for text. You can open up chat on the side. Go ahead and submit questions or comments through that and feel free to address everybody or specifically one of the presenters if you care to. I and some of the other folks hosting this webcast will be keeping an eye on that. You can also raise your hand throughout the presentation and you'll see that over in one of the side panels as well and we are going to be recording this webcast, participating in the webcast notes, your consent in the recording. The webcourtings and presentations will be posted to both every and Stanford sites and follow-up and with that, wanted to note that we have had this ongoing series of webcasts. Hopefully you've been able to join us and if not, hopefully for additional future ones co-hosted by Eprion Stanford and specifically the Bits and Watch Initiative at Stanford. So really excited to be here today with a couple of representatives from the academic space and presenting the academic panel. Excuse me. Through this summer webinar series, we've been working to convene experts across disciplines and really all of those that are related to the integrated digital grid to exchange our views, to identify gaps and really understand what we need in order to bridge gaps and move forward together in this space. Ultimately, there is an opportunity for a research roadmap and a collaborative initiative where we keep all of these stakeholders and keep bringing additional folks to the table to again, try to weave together all of the expertise and research needed across all of the categories that you'll see on the right hand part of your screen. Really an exciting endeavor and a really important one and it's likely one that you all have been on for the last 10, 20, 30 years, but moving towards a truly integrated grid and moving beyond just integrating the more traditional pieces of the grid and moving to the grid edge as well. And we do talk about this vision where we really look at those grid edge resources, customer resources, distributed energy resources, distribution connected resources and look at how we can really operationalize all of those resources and come to a low cost solution and optimize grid flexibility. In order to support this, we have additional webinar series coming up on the 22nd of July, next Wednesday, we'll have the government panel and then finally on July 29th, the final planned webcast on corporate research and again, remind that you still have about four minutes left on the pool to let us know what you're interested in seeing at future webcast then you see all of those opportunities up here, all of these closely related to the shared integrated grid future. So definitely give some thought to both long-term strategic needs as well as more immediate needs and topics that would really help you get stuff going both today, tomorrow and then help prepare you for 10, 20 years out. So continue to give that thought or go ahead and chime in and complete that poll. And then finally I'll wrap up and introduce our speakers and get to the meat of this discussion. Really excited and honored to be joined today by two representatives from close partners of ours in the shared integrated grid space. We've got Dr. Ram Rajagapal from Stanford University, the director of Stanford Sustainable System Lab and he is going to be talking to us today about looking beyond technical considerations and integrating distributed energy resources and talk a little bit about learnings that they've had through projects on practical constraints and talk a little bit about accounting for physical and cyber constraints, privacy, equity and talk about some of the next stage research challenges in an upcoming project as well. And our other partner that we have with us today is Dr. Amril Fareed from Dartmouth University and he is the director of the Laboratory for Intelligent Integrated Networks of Engineering Systems or the Lines Lab and Dr. Fareed will be talking about, again, the beyond technical challenges, simply understanding how stuff operates, how reliable it is, what it looks like on its own is one thing, but we'll hear a little bit about the need for an extensive, extensible information model, common information model for energy and internet of things devices. What do we need in order to support not just ubiquitous connectivity, but the information exchange and talking that needs to go on between devices in that space and how that can then support the economic benefits from that grid edge flexibility being operationalized. So thank you both for joining us today. We will be hearing from each of them for about 25 or 30 minutes and following that we will open up for discussion. If you have questions, particularly clarifying questions throughout the presentations, please don't hesitate to chime in through chat. I'll be keeping an eye on that and look forward to hearing a little bit about how universities play a really vital role in moving this space forward. And with that, let me just make sure I'm not missing anything and we'll move straight into hearing from Dr. Rajagopal from Stanford and I will turn it over to you. Welcome, Ram. Hi, Sarah, let me turn on my camera here. First of all, thank you for hosting our session. I also wanted to thank Epri and the Stanford Bits and Watts program for inviting me for this event. Just here on the left-hand side of this very first slide, I chose a picture of something that is hot off our research to show the power of open data. We have been able to analyze the demand, hourly demand from 58 countries around the world to understand the impact of COVID-19 restrictions on the demand. And you can see here, this work was entirely done by a group of students and postdocs. So I thought a little bit about what is interesting today and kind of the title of the presentation reflects that. At Stanford, what is our, some of our key interests? And these are the concepts I have here, virtualization, learning, coordination and privacy slash fairness in these consumer-driven distributed energy systems. This is very important because the word consumer here means there is people and their preferences that need to be taken into account. So I think that's a major distinction of the traditional view of such systems. So let me just see if I can. So, you know, we all are familiar with the grid. I'm just gonna go very quickly here. And the traditional grid is changing because we have massive adoption of renewables and new technology such as storage at the transmission level, but at the same time behind the meter, consumers that are adopting new technologies as well, like EV charging, smart appliances, self-generation through solar PV, for example. And all of these technologies are connected to the cloud. This enables us to collect data and send signals and potentially coordinate all these resources to achieve individual and joint goals that can serve, you know, to the wholesale market. So you can have flexible load that follows generation instead of the other way around. But also as we start to see increased adoption of these technologies, this coordination can be used to mitigate the impacts on the distribution network. So one of the research focuses here in Stanford has really been around how do we utilize this capability behind the meter to support the rest of the grid ecosystem? And when we say that, what are the key challenges here? The first one is that we need to understand what is the real flexibility available at each customer? And that critically depends not just on the technologies they have, but also their preferences. How much are they willing to change, adapt, and adjust? And learning that at the scale that needs to be done is something that still remains extremely challenging and an open question. And something that I feel, for example, you need not just methods, but a lot of open data interaction with customers and so on. The second question is, well, if we know that consumer preferences and assuming we have a model for the distribution network, then we can start to think about how to scale this coordination to millions of heterogeneous resources. Again, the keyword here is scaling. There is many, many strategies and architectures for coordination. This is very well known, but how do they actually scale to the scale we need them to go? The third one, I think is very critical and goes into the realm of economics and policy because it is all about whatever strategies you create, whatever platforms, ideas, et cetera, be it to collect data or to coordinate, you definitely need to make it easy and attractive for consumers to engage. That means not asking them tons of questions or asking a lot of effort on their part and also mechanisms to compensate them. Last but not least, I think we are seeing a dramatic change in society today regarding two ideas. And I think it's time we took them more deeply into how we design these energy systems, particularly consumer-driven ones. One is this notion of privacy. We are very familiar with this. There was some early work on this looking at the privacy of smart meter data and so on, but in those days, the entity that took care of the data was just the utility. So the privacy was in a way managed by utility regulators. But now we are talking about aggregators, different types of service providers and so on. We need a broader and expanded definition of what privacy is and actually ways to enforce it, monitor it, measure it. And the second topic I think is one of fairness. So I will address those briefly at the end of the presentation. What we mean by fairness is we need to give both equal access to the participation in these modern systems, but we also need ways to ensure that the systems we create don't exploit particularly vulnerable customers. So families who have lots of people who stay home tend to be lower income and maybe your demand response algorithm would continuously target them for turning on and off. These are considerations that have been taken into account in economics and more now on web search and things like that, but I think there's a whole open question about how to do this from a technical perspective, but also institutionally, how utilities and companies, aggregators, technology providers, policy makers are gonna deal with these issues. So sorry, it takes me a little time here to change the slide. I just wanted to kick off the conversation with something that we did recently which is kind of unusual here for our group. We worked with a sociologist and a psychologist to try to understand in a real population how much actual consumer flexibility there is on their loads and devices that you see. And in this graph here, what you see is really what percentage of this, each one of these activities is performed during the peak time. This is a population of 370 individuals and the paper came out an energy policy around this. And you can see here, for example, TV is watched 90% during the peak time, et cetera. And in black, that's kind of the willingness to shift what percentage of people were willing to shift. And you can see that just very few items like your washer, dryer, dishwasher are things that both meet somewhat being used in the peak time and somewhat willing to shift. So this started shaping a lot of our groups thinking on how to design the systems that respond to this need. And I'll address that when we talk a little bit about PowerNet. But I think this kind of engagement and interaction with consumers, now they're doing a project where we educate these households through their kids to understand their own consumption from smart meter data and then try to execute on the shift. It's a little bit away from the traditional path, but it's all about how do you have this human in the loop during this kind of management. And it's a ton of work and very interesting as well. So if you come here in the summer, we have workshops with hundreds of kids and it's super exciting. Okay, so going more towards learning, the first idea I feel is very critical is that in economics, there's a gold standard about how you learn preferences. You usually say, people have a utility function. I'm gonna show them different choices, observe the choice they make, and then model this utility function that is kind of the embedded value of the choice. And there is lots of techniques for that and it's well known. This typically translates in the utility world of electric utilities as we will do randomized trials on different programs that we have, let's say for demand response to understand the size of the effect and then model the heterogeneity of that effect as a function of customer characteristics collected from questionnaires. It turns out that these models in the electricity side perform not so great. Typically, if you created models like this and we work closely, for example, with PG&E on looking at this, and then you say, now I wanna recruit enough customers to have a megawatt of demand response on this particular feeder, you end up with something like 10 to 20% of that. That's kind of called the yield. Just as a point of comparison, if you go and talk to somebody at Amazon, they will say that these offers that are put on your first landing page of the Amazon website, when you see it, 75% of the people click in one of those offers, and then of those clicks between 50 to 70% of the people actually purchase what they saw. So that is overall much more than like a demand response offer where you're actually offering money to the customer and yet the engagement is less. Economists looked at these questions in the last 10 years and there's a really beautiful stream of literature on that. And what they kind of found is that this is called the issue with declared preferences, stated preferences, because normally we are not able to observe what actions customers take with regards to their electricity consumption. So you do polls and surveys like the one I had before and then use those polls and surveys to predict how much flexibility customers have and that goes directly into the planning. And so those predictions don't completely capture the preferences because customers don't pay a lot of attention to their electricity bill. It's not a big part of their budget. So this is an issue not just with electricity in many other areas. Declared preferences of things that don't weigh on your pocket as much are not accurate. So one idea that we developed in partnership with PG and ESE and a host of utilities was this idea that, well, I have a smart meter. That's like high resolution temporal data. That data, it's like an MRI of customers' preferences. If I could interpret that MRI and by that here is what I mean extract features like what's the peak load, base load? What's the thermal sensitivity? What are the load shapes? I can use these features to build these models for segmentation, targeting, et cetera. This gave rise to a project called Wisdom which was carried out here at Stanford from 2011 to 2016. And then we released this open source package called Wisdom. It's now used in eight U.S. utilities in Southern California Edison, PG&E, Vermont, et cetera. And then a few more worldwide. And it's completely free open source. You can download and I think in 2016 when we released it we had analyzed about eight million smart meters with this package. Since then, we haven't really kept track of what's going on. So just as an initial, like the very first result from Wisdom is shown here, we just did the first study ever at the time at the scale that was done which was 66 million customer days. We looked at 66 million customer days across the Bay Area and performed clustering. And when we started doing this study PG&E gave us the data that took itself like a year or two. And then when we did the study initially, the consensus was, well, why should we do this? It's well known in planning that all customers are these dual peak shape. And then when we actually did the clustering we found many different load shapes for each day including just consuming at night, very flat consumption, evening peak. And the so-called popular or planning option it's only 14% of the load shapes. And this is very critical because for example load shape number 811 here actually can have demand response capability at 3 p.m. Versus this guy here 14 and I'm hoping you guys are seeing my mouse can have no such capability or no such significant capability. So first lesson from this project is when you're thinking about this new kinds of flexibility that are emerging, number one, due to the preferences of consumers the flexibility may or may not be there like you saw from the survey. Number two, data and massive data are essential to have accurate models of consumer preferences to achieve the types of performance we need to make these things cost effective for the future. Okay, so the second question we had in our list up there was okay, so we understand customer preferences maybe we can collect a few of these consumers and use their capabilities to offer services back to the grid or to reduce their bill or something like that. How do we do that at scale? So that was the genesis of this project PowerNet. It's a project that has been funded by the California Energy Commission and RPE as part of the nodes program. And we had as partners the SODN, Suntech Drive and Google. Google also funded the project and it's also a partner of the technology. And it started out with a simple kind of interesting observation. At the time we started this project in my group and in Abbas El Gamal, he's another faculty here at Stanford in his group, we had had a few collaborations around optimization algorithms for storage and kind of distributed control. And this is kind of a big fad in power systems. And these algorithms work great in MATLAB and that means in simulation. And then I visited Google and Arun Majumdar's former group, the one that he created because I had this kind of question that was put forward by my student Gustavo. Well, but how do we know these algorithms actually work in the real world? And then somebody told me Arun was doing this in Google and we went and visited that group. And what they had tried was actually to implement in a lab scale with realistic constraints in terms of the cyber constraints, the information exchange constraints, several of these algorithmic strategies. And what they found out is that fully distributed algorithms, although very elegant and beautiful theoretically, they don't really scale well. And it's very hard to diagnose, it's very expensive to try to set them up with customers and so on. And in fact, at that time, they had just managed to do like one home and technologies inside it and tried to coordinate the energy consumption. And they had done a survey with customers because they needed to collect preferences and Google does these user surveys and they said, well, you're gonna collect your preferences. And here's our idea, we're gonna put forward a questionnaire with 50 questions that you need to answer. So we know everything we need to know about how you like your house to be, like your temperatures at point and so on, so forth. And the result of the survey was that the immense majority of people said they would never use a system like that because I just thought about my own home, there is three preferences for the thermostat set point and nobody can really agree on it. So that's an example of, trying to do that 50 times is impossible. So we decided when we went to look at nodes, we wanted to propose a project which was about building a real system that can do such coordination, which addresses the scaling issue. And that means really addressing three critical problems at the same time. One problem is whatever coordination mechanism you create, its coordination is gonna happen also through the power network. In our case here, we focus on the distribution network. The second is whatever mechanism you create needs to learn and understand consumer preferences. In our case, we picked the consumer to be homes initially. And the preferences of the homeowner have to be taken into account into the coordination. The idea of picking up homes was also because each home by itself cannot really provide much value back to the grid, but in coordination, they can do a lot. And the third issue is that whatever scheme you create or architecture or solution you create, needs to be able to account for the real constraints out there in terms of information exchange, so data exchange and communications and so on. We spotted at the time this notion of, well, we can use the cloud to do coordination. And for example, this is already done in data centers. So the cloud runs the coordination of the communication between the server racks. But the key idea, even in the data center case, is the cloud cannot do this in real time. Because whenever you put the cloud in a feedback loop, there is delays on the information exchange. There's reliability issues. If you try to make it geographically spread, and this is all stuff we learned from Google. So how do you do it so that the cloud is not really at every second and every minute calculating and sending the decisions to all the homes? So that was the origin of this project, PowerNet. And here's how the system that we came up with was laid out. And again, we started this project in kind of an unusual way because we decided to build a lab in Stanford, put these devices and things together, work with Google, connect everything to the cloud, et cetera, without knowing what the solution was gonna be. And we experimented with tons of algorithms until we figured something that worked and that also had some theoretical justification. But that is still kind of work in progress, the theory around this. So the key idea of the system is that every home has a smart hub, a local intelligence, it's a computing capability that communicates digitally with your inverters for the solar panel, storage, your EV chargers, and also with this smart dimmer. This is a new kind of smart panel that was designed by our colleague, Professor Juan Rivas, working with students. This panel allows you to have high resolution measurements of all of the sub-circuits of the home, but also the capability to do voltage control at the sub-circuit level. So you can have some controllability of loads, appliances, and so on, even without actually the appliance being smart, necessarily. So that's kind of the resources you have. And normally this home would have a bill and it tries to minimize the bill using data and these resources. But in order to minimize the bill and also to do coordination, which is homes might get together and say, well, let's shave the peak in this transformer because we'll get some compensation from the wholesale market or maybe directly from the utilities and from the demand response program. In order to do all of that, they do need to understand an account for the impact they will have on transformer capacities and network voltages. So that's kind of the setting. And how did we solve this? So I'm not gonna enter into a lot of details, but I'll go a little bit into this project. First, it required thinking about this architecture of how does the local controllers and this kind of cloud coordinator engage? And we also imposed ourselves this issue that we wanna minimize the information exchange and the cloud cannot run in real time. And so what we found is an architecture that has a planning stage that's run every 24 hours in alignment with the day ahead market. And in that planning stage, it sends the dynamic power bounds. So it's like establishing a bandwidth for each one of the homes in terms of each hour, how much power are they allowed to draw or inject back into the grid. So instead of having a static bound given by the transformer, they can have a dynamic bound. And it turns out if you do these calculations cleverly, this is enough to really protect the network while enabling the coordination. We designed this smart dimmer, which I'll have, which I can, if you want things open up, you guys can come in and visit here at Stanford Lab. We also did lots of algorithms with data from predictions to learning the customer preferences, from observing their choices and using appliances. And then the biggest effort in the project was a massive kind of deployment. So we had a large scale lab that was kind of built between Stanford and Slack. And then we have deployments in a real world farm and 20 homes in Fremont, where we put battery storage, solar, we read their data, we kind of do all kinds of things. So I'm just gonna point out one thing here and then move on. The critical challenge on the coordination is that there is a spatial and temporal data symmetry. It has not to do just with the technical issues. It's because who knows what information. The utility knows the network, but doesn't necessarily have access to behind and shitter devices. DR providers have access to their own devices, but no information on the overall load are the utility network and so on. So how do you deal with this data symmetry? And so the design of PowerNet addresses this issue by figuring out how the information needs to be shared and how that shared information is utilized so that we can maximize the benefits for the consumers. There are actual participants in these systems, sorry. So we initially, when we designed this system, one of the things that we learned and I think this is kind of the first major lesson is seeking this kind of elusive optimality that you see when you design algorithms and you test them in simulations. It's actually detrimental in designing real scalable solutions. What we found is by sacrificing some of that optimality, you can really take into account all of these practical constraints and achieve good results. So you can reduce the voltage deviations that would result from people synchronizing all their decisions. You can also do things like offer ramping services, regulation services, we showed all of those. And then if you look at the end of the month into their bill, you still achieve like 90% of the ideal oracle reduction. So they can't even achieve that in practice, but just that small sacrifice results in big benefits. So here's kind of the lab that we built and you can see this is a lab that was a partnership between industry, the Stanford School of Engineering and RPA-E. So in industry, the engagement was through this bits and watts program that Liang direct. And you can see here, we have like a smart home with real appliances. It's connected to a panel. It's connected to a solar panel, to a battery system, water heater, the whole thing and to an EV charger. This home is connected to a home that's a little bit more virtual. And here's like a grid emulator. All of this system is connected to something like 10,000 virtual homes in grid lab D and open DSS now. And this whole lab is connected to a similar lab in Slack that had more of larger loads and were meant to emulate small and medium businesses. And this is where we initially tried to design our solutions and demonstrate them. And we learned so much from, first of all, interoperability. Yes, we do have the SunSpec Alliance, but if you try to go and read any information from real inverters, it doesn't really work. Okay, so just to continue here, we did the field test and we have successfully deployed this system in a farm. So for example, and then now we are doing it in homes in Fremont, you can see here that we are controlling the cooling of cows in this real-day refusility. This is one of the largest day refusilities in the United States. And they spend about $200,000 every summer with this cooling. And what we are able to show is that using PowerNet, you can reduce about 50% of that cost. And what PowerNet does is it coordinates the storage, the solar panels on the rooftop of the barns and the fan control and tries to maintain the cow temperatures appropriately. Right now using airflow measurements, but we have installed like a thermal camera. So as you can see in system, I use this as an example because there's a lot of embedding of IoT solutions from different vendors and so on. So I'm just going to wrap up with our latest project and maybe just need a minute for that. So after we did PowerNet, we realized there are a lot of principles on how you should design the software solution for this and the architecture. And we wanted to look into how to do this at an even larger scale and we partnered up with folks in consumer systems design here at Stanford and decided to look into this issue of virtualization, modularity and layering, which is their principles of how they do computer systems design. And that gave source to this project called TrustDR. The goal is to ensure resource virtualization, security and privacy for all of these algorithms. By virtualization, it's a simple concept. Can you take one resource and make it look like many virtual resources or many resources and looking as one from a software perspective? So when you write your solutions, you don't have to worry about this. And then in the privacy and fairness, it's all about really protecting the privacy of the customer. If I have to have data privacy, how do I design algorithms? It turns out to be pretty difficult to do that. To conclude, I just wanted to leave the form of engagement we have had through all these projects is using this affiliate model for partnerships through the Bits and Wads program. We work very closely with companies gathering their inputs, having them participate in board of advisors and in some cases directly join our research teams and students who finish this project sometimes go on to develop these technologies commercially. Sometimes they go into these companies as employees and so on so forth. Also have a flagship program that was started through this partnership around EVs. And if you're interested in it, please contact me. Thank you very much. Excellent. Thank you, Rom, appreciated that discussion. And did want to thank folks for pushing questions through Q and A. Please feel free to use that or chat. We're keeping an eye on both. And as we wrap up, we do have one question coming in to start off discussion after Rom has finished talking. Thanks again, Rom. And now we're going to open up for Amral to give us a discussion on the work going on at Lines. And you should be all set, Amral. Take it away. Yeah, thank you, Sarah. Today I'd like to be talking to you about accelerating the shared integrated grid through an EIOT-extensible information model. And there's a lot to that. So let me try to unpack it. Really what we want to do today is conceptualize the development of an Energy Internet of Things extensible information model, building upon the collaboration that the Dartmouth Lines has had with APRI. So first we have to start out with the Energy Internet of Things. And I have to thank Rom for describing a lot of these different energy Internet of Things devices. It's really all about appliances and different forms of energy resources that are network-enabled. And what you want to do is make sure that they end up participating as part of a shared economy. And what we find is that these Energy Internet of Things are actually going to be quite ubiquitous. You can imagine every single appliance in our home is going to be a smart appliance. We all know this when we go into the stores or Amazon and so forth. But it's not just going to be in your home. It's going to be in commercial businesses and industry and throughout the grid as well in distribution system, transmission system as well. And a quick plug of our book where we have talked about what is the Energy Internet of Things and how does it help support the sustainable energy transition? This is an open access book. So if you search for it online, you'll be able to get a hold of the book very easily. So that brings you to the question of, well, what is an EIOT extensible information model? Well, if we're going to be all technical about it, then really it's an extensible collection of nouns and attributes that provide a common language for describing EIOT devices and how they communicate with each other on the internet. But really the reason why this is so important is that if you're going to have all of these different devices talking to each other in different ways, you're going to need to create a common language, a lingua franca for all of these devices and that also engages people as well. So why is EIOT so important? Well, we've been working in the lines in the area of renewable energy integration for about a decade now. And again, as Ram mentioned, the grid is changing and in particular it's changing because we're integrating solar and wind. These are variable renewable energy resources. And at the end of the day, they end up eroding the capacity factor of dispatchable generation units. And so now what you're going to need is new dispatchable resources that you can end up controlling and you really don't have much place to go than on the demand side through these EIOT devices. And the majority of these EIOT devices are connected to the distribution system at the grid's periphery. And so that means that we're going to have millions, if not billions of heterogeneous devices and we're going to have to try to find ways to coordinate them together. And this is honestly a rather big challenge. We need to have the right algorithms in place and they need to scale. So in order to make it work, you need a not just a common information model and EIOT, but you're also going to have to create what we're going to call a shared integrated grid. And that means that customers are now an integral part of the solution there. We can no longer have a paternalistic grid, if you will. So the shared integrated grid is not just about EIOT, it really must have two other really important pillars. One is customer engagement in that customers need to feel like they are part of the solution in engaging and that the grid is providing services to them and potentially also that customers are providing services to the grid. And then of course, you also need to have community level coordination and achieving that coordination is in many ways easier said than done, but we'll talk more about how that might begin to happen. So in order to make it work, you're going to have to have a collaboration platform in some way or another. And one of the things I would like to do in this presentation is highlight some of the efforts that we've been doing in New Hampshire in a very practical way with the various energy and grid stakeholders in this state to try to make collaboration such an important part of the state's energy future. If you look carefully, you're going to find that there are a number of different collaboration platforms that are popping up everywhere. And one of them is the City of Lebanon's Energy Advisory Committee and they have been leading the effort when it comes to community power aggregation in New Hampshire. That's ongoing. Where Dartmouth is, the Sustainable Hanover Committee has been implementing pass-throughs of wholesale real-time pricing to municipal buildings and because wholesale electricity prices are on average, even though they are time-varying, on average they are lower than retail rates that have saved a lot of money for the town of Hanover. There's also a New Hampshire Community Power Coalition that is bringing together New Hampshire cities, towns and counties to try and implement community power aggregation for a wide variety of localities in the state of New Hampshire. And then there are two efforts that are going on simultaneously within the Public Utilities Commission. One is to support the actual implementation of community power aggregation. This was a new law that was passed in 2019 and that's ongoing this year. And then also New Hampshire is well on its way to standing up a data platform that will allow for customers to share their data with a wide variety of stakeholders. And we'll talk a little bit more about that in a few slides. But honestly, it's not just in New Hampshire, certainly not. There are a whole bunch of collaboration platforms that are really popping up all over the world. And one of the ones that I just wanted to highlight is Elia and the Internet of Energy that they are doing in Europe. But one thing is for sure is that if you are going to try to get coordination across the distribution system and behind the meter, you're definitely gonna need collaboration platforms to make that happen. And you're gonna populate the collaboration platforms with participants. So you're gonna need an actual consortium. So let me come back to this notion of community power in New Hampshire. And I mentioned community power in New Hampshire. And it has several municipal members that are working together and then a number of support members like Clean Energy in New Hampshire, which is a nonprofit. We are supporting them as well and Community Choice Partners, which is a consultancy. If you look carefully as to all of these participating communities that are interested in community power aggregation, you'll see that it makes up about 13% of the New Hampshire market just off the bat. So there's a pretty strong engagement just from here on in and we expect that that will only grow with time. As for the New Hampshire Energy Data Platform, which is an ongoing docket in the state of New Hampshire, there are quite a few different interveners. And I think this has captured the interest of a lot of different grid stakeholders, both in the state of New Hampshire and beyond. So, you have representatives from state and local governments. Of course, all of the utilities are involved, academia, industry experts and small businesses, startups, non-for-profits, vendors and legal counsel. There's a lot of different issues that need to be sorted that are of course not just of a technical nature, but also the economics, the value of the data, the privacy of the data, the governance of the data platform. And so you would expect to be bringing together a wide variety of different players in order to make all of this work. So how do you set up a common information model? And probably the best way that I can talk about this is not about the common information model itself, but maybe the data platform that it would end up implementing in the end. And so what I have here is just a conceptual representation of what the data platform might look like. And it's gonna have all sorts of different stakeholders connecting to it, whether they are governmental agencies, academia, consumer, independent system operators, load service entities, maybe curtailment service providers, the distribution companies as well, and maybe authorized third parties. And they're all gonna be exchanging data in different ways to support their activities. And so you would probably wanna ask yourself, how might we think about building such an energy data platform and what are we going to have to pay special attention to? So I'm gonna give you about 30 seconds to think about what the answer is. And if you can write your answer into the chat, I'll build that into the rest of my conversation here. One answer is just start coding. And another answer might be write a request with proposals and outsource it to the load critic. So did I get any answers into the chat? Are we all awake? We've been talking at you for the last 53 minutes. Okay, well, I would say that the first thing you really need to be thinking about is that you have a wide variety of stakeholders. It's all about the people and all of them have different needs for what this data platform is going to do. So you have to identify the stakeholders, the requirements, the potential use cases and that they all have different places with respect to the legislation, the regulations and their various needs and you're gonna have to collect all of that together. And one of the things that you really wanna make sure is that if this data platform is effectively going to become a public good, a public infrastructure, whether it's privatized or not is an entirely different question, you have to make sure that there is a place on that platform for everyone. And it's also going to have to make sure to respect the local laws. And so last year we had SB 284 and SB 286 passed Senate bills that is passed in the state of New Hampshire. One was for the data platform itself, but another one was for the community power aggregator. And for community power aggregation, it provided different authorities to community power aggregators that honestly are amongst the most progressive authorities that we will find in the country, if not the most. And these include the supply of electric power, demand side management, conservation, meter reading is a big one and potentially a contentious one with some utilities and other related services. And if you're going to end up providing all of these different authorities, then that means that there's needs to be data that supports those authorities actually being enabled by the community power aggregator. And so you wanna take that into account in the design of the data platform. Then you have to go through what systems engineers would call a requirements engineering process. You have to reconcile all the different requirements. You have to make sure that they are turned into technical requirements. And you also have to make sure that they're all equally valid. So when I mean equally valid, imagine building a hypothetical road. Well, if you were to listen to a pedestrian, a cyclist and a motorist, all of them were gonna say different things. I wanna sidewalk a bike lane and two lanes. And you have to make sure that the road meets all of those different requirements at the same time. The data platform needs to do the same. And then technical requirements is completely different. For engineers, we often think of technical requirements because of that's our training. If I want to control the comfort of my home, I would probably say something like 72 degrees Fahrenheit and 50% humidity. But if I were talking to the, maybe the other people in my family, they would say, can you just make it warm and cozy? So there's a translation from customer requirements or stakeholder requirements to technical requirements. And that needs to happen in order for a data platform and it's underlying extensible information model to happen. So part of the reconciliation means that you're gonna have to end up classifying all the different requirements. And that's not an easy thing. There are gonna be requirements related to operations. For example, maybe you wanna help implement a transactive energy market, more on that to come. Or you wanna help operations improvement, maybe in the case of energy efficiency monitoring or improvements, maintenance, like scheduling the maintenance of a motor. Maybe we can see the power quality changing as a function of time or maybe even supporting regulatory compliance. So in New Hampshire, we have what's called the 900 net metering rules. Yes, there are 900 of them. So you can imagine the data requirements necessary to actually achieve compliance with that and also help regulators see that those rules are actually being met. And you can also think about another classification here. This is straight out of a textbook. If you were to look at system requirements, they have all these different types and I often teach the differences between these types of requirements in my systems engineering class. So once you have all the different requirements and you turn them into functions that you want the extensible information model data platform to end up having, then you can begin to think about quantifying the benefits. From function arises benefits. And then you can start thinking about, well, what's the relevant data that is going to help actually achieve this function with a focus on interoperability and extensibility? So here I have focused on the extensible information model. The IEC for a long time has had the common information model to address different aspects of the grid. In order for us to truly integrate EIOT to the degree that we want to, then we're going to have to build upon that to include all of these millions of devices and smart appliances that we hope to have a coordinating function in the grid. And as we think about all of that data, that will end up bringing about the cost. So the system form drives the cost of the system. And then the last thing you want to think about and certainly not the least by any stretch is you have to address governance and implementation challenges. So I'm going to ask you to once again join me on the chat. What do you think might be some important governance and implementation challenges? The Jeopardy theme song is not working this time around. That's okay. So my first answer is, ah, no problem with governance, we got this, what could possibly go wrong? Actually quite a bit can go wrong because naturally you have a wide diversity of stakeholders that all have different needs, customer preferences, privacy, all needs to be brought into bear. You have to ask yourself, who is going to end up implementing the data platform and then who's going to end up operating the data platform? We have to recognize a fundamental point about data and economics, which is that he who has the most data in a market ends up having the most market power. And so that drives a lot of the questions relating to governance in one way or another. You can imagine if you had all of the data to the New York Stock Exchange, let's say before the transactions had actually happened, how much power you would actually have. And we are talking about something of similar scope, although in terms of or similar scope or similar importance. So once you have thought carefully about an extensible information model and maybe the data platform, I would like to highlight what you could potentially do. And we've had a project going on for several months now in terms of standing up a transactive energy blockchain prototype for the city of Lebanon in collaboration with the city of Lebanon's efforts to develop a community power aggregator and also with Liberty Utilities. So all of us know the conventional power grid that has our supply chain of generation transmission distribution and consumers. And it has a generation company, a wholesale market and a utility and we buy our electricity from the utility. Now, if we start talking about having municipal or community power aggregators, the consumer is going to end up buying directly from the aggregator. And by the way, in the state of New Hampshire, this is an opt out law. So now the municipal aggregator is the default provider of electricity, not the distribution utility. And then the municipal aggregator can buy directly from the distribution utility or it can buy from the wholesale market as well. So that means that the aggregator has the potential to actually stand up its own distribution system level market or an aggregator that has a market that's built in. And so we have been working on the transactive energy algorithms that would help enable such an aggregator to do real-time pricing and end up providing to consumers a lot of choice, maybe greater access to clean and cheaper electricity, giving them access to real-time wholesale prices and try to enable peer-to-peer electricity trading. And naturally a lot of data is required to make this work and we're developing the software as we go along. So you might have heard that there are other transactive energy companies out there, Power Ledger and LO3 Energy are some of the ones that come to mind. And as an academic who's been working in the power system space for some time, I have to ask if there are indeed guarantees of convergence and physical security by these different mechanisms, they don't tend to emphasize those very nuts and boltsy technical things in their information, but we are working on an algorithm that is based upon the ADMM and it guarantees convergence, physical security and economic optimality. And we've been collecting data from Liberty Utilities here. I have a picture of the 10 or so feeders that are in the city of Lebanon, a city of about 15,000 inhabitants. And as I mentioned before, the city of Lebanon is leading the way in the state to stand up a community power aggregator and they're very interested in having that community power aggregator provide a transactive energy service as one of the services available to its inhabitants. And again, they're very interested in cheaper electricity for its inhabitants, but also enabling greater penetration of renewables as well. So as you can see, if you put all of these different pieces together, you can think very carefully about enabling a shared integrated grid and those pieces are the EIOT itself, the network-enabled devices. And then of course, engaging the customer and community level coordination. And I hope I've given you a sense of how these three pieces can end up coming together with the various initiatives that have been happening here in the state of New Hampshire. And I think that's it for me today, Sarah, until we have questions, that is. Fantastic. Excellent. Thank you, Amro. Thank you, Ron, again, for sharing with us some of your work. We have had some questions come through Q&A. And I'm going to start a little high level and then we'll dig in a little bit deeper. If you both don't care to give some consideration, Ron and Amro, to how DER measures and grid digitalization affect grid resilience. Let's take a couple of minutes for that question. Ron, would you like to go first? Either way, Amro, very nice presentation, by the way. I think it's a very difficult topic, the creation of the actual platforms. So in terms of the resiliency, I think there is two key ideas. One is understanding how do you translate consumer flexibility into resiliency? Because what is resiliency? How do you measure it? For example, if I have an earthquake in San Mateo, it will affect the homes. It might disrupt some of the solar PV. It might affect part of the grid. What do I mean by being resilient? Do I have to provide full power to the remaining homes or is it part power? Does it depend on the intensity of the event? We collaborated with the faculty who does earthquake engineering to try to define these metrics around DERs. And it turns out there's two key aspects in my mind. One is you need to understand something like what is the full load potential, half load potential, no load potential of each customer. The second one, I think it's a lot harder, which is really any system to be resilient beyond its individual capability, there is this notion of a network resiliency. And that means you need to understand how to model the switching in the distribution systems and adding those switches, as well as how does damage due to different events affect those systems. I think it's fairly open question. From a broader utility perspective, I think there is yet another question. Once you have a resiliency model, you need to know what resources are out there today in terms of DERs, from solar panels to different equipment. And the mapping of this has been lagging tremendously. This data is hard to access, often unavailable. In Stanford, we started a project called the Stanford Energy Atlas that uses the satellite imagery. We work with people in machine learning and so on and do country-wide mapping of these resources. But there's many such efforts now that are happening. And I think without that, it's impossible to do this resiliency assessments. Yeah, so the question of grid resilience is one that is particularly present now. And we can really take it in a lot of different ways. So the first thing that I really would like to emphasize is that grid resiliency and grid sustainability go hand in hand. As we start to think about integrating batteries and integrating solar PV, particularly in the distribution system, the presence of additional energy resources and maybe even meshed lines from a structural perspective has a huge improving effect, positive effect on grid resilience. And we have a new paper on that that has done that study on the scale of the United States as a whole. Then once you get past the distributed energy resources themselves, then there's the question of measurement, monitoring and potentially control. And there, I would say digitization can only have a positive effect in the long run. And that's because a lot of the distribution system and also what's going on behind the meter is not really visible to us today. And if we increase that visibility, the likelihood that we will be able to take effective action in the event of various forms of disruption is also going to improve. Now, does that mean that the problem is solved? All we have to do is just put lots of metal and wires and devices onto the grid and we'll be all set. The answer is clearly no. There's still a lot of work to do to achieve the right types of control mechanisms, the right types of markets, the right types of regulation and also the customer engagement to get the resilience that we're looking for. And I'll conclude by saying that here in New Hampshire, there is a clear independent streak amongst our residents. So resilience is a very important part of the culture of New Hampshire from an energy perspective, but it's also very much tied to the environmental streak that is here as well. Great, appreciate those responses. Another question that came through and this is maybe more directly related to where you wrapped up Rom, but in the Q&A box, we had questions about the time resolution of cloud coordination. What is the time resolution? I don't know if you want to dig into a little bit about that, also about the optimization model and how you deal with changing household preferences and your unmute. I answered it on the text there, but overall the ideas in the cloud coordinator itself, we run it five to 15 minutes, so when you're offering ramping services, et cetera, but locally, we run it between a minute and a second. We have tried all of those combinations. The minute seems to be more than enough. The optimization, we started with SDP, conic optimization to take into account the network. Obviously required a lot of magic, make it work with like tens of thousands of homes and large feeders, but recently we moved to the idea of successive linearization, which despite the absence of strict performance guarantees, in all of our practical examples and real feeder systems that we got from a few utility partners, it has done a really amazing job and it's super fast. On our side, the transactive energy algorithms that we're looking at, which would be through the cloud at the end of the day or through the broad internet, pretty much the same on the five to 10 minute time scale. And there's still a real driver for that because you do wanna synchronize with real-time energy markets, but for actual low level control or control, let's say within a home, then you could actually be thinking at a faster time scale around a minute. It's at the end of the day, pseudo steady state control. So a minute is a really nice time step. Excellent. Appreciate that feedback guys and the response and the Q and A column as well. The final question, actually before we move into a couple of even higher level questions that I have for the speakers, any additional questions that people have and would like to submit orally, please feel free to raise your hand in the webcast. Tara, I think there is a question from Larissa on the chat, the Q and A chat. That was great. Oh, here we go, great, perfect. Yes, so another question came through on the potential for developing local energy markets, distribution operational markets as distinct from wholesale market opportunities and how can those distinctive opportunities be optimized? Yeah, so when we think about wholesale electricity markets, they really pertain to the transmission system. But there's this whole amazing world of distribution systems that go largely uncoordinated and where so many of us everyday people have a chance to affect the on the ground reality of distribution systems. So I often like to give the example of, I am someone who works at home and maybe uses heavily electricity during the middle of the day and my neighbor is not at home, but he might have solar panels. And why isn't that someone who is maybe only 50 or 100 yards away, I can't buy electricity from him rather than through this rather pardon the pun, but circuitous route through the wholesale market and my utility and so forth. And if we really start thinking about trying to have the grid be a place where community structures end up revealing themselves, then I think that there's a real opportunity to improve the way the distribution system works in much the same way that we think about improving our local schools and communities. Again, here in rural America, community structures are a very important part of making local government work and towns and cities work. So those are just some first thoughts on that topic. I think to add to what Amuro said, if you look not just today, but if you're looking at the same grid 10 years from now, for example, you'll have much deeper penetration of EVs, particularly kind of delivery vehicles and things like that that pull a lot of power and you will need to start managing the impact of this power on the distribution network much more intelligently than today. Today there is a big slack capacity on the distribution system. So we are taking advantage of that. If you go beyond the United States and look at other countries, for example, in India or in Brazil, distribution transformers are often already overloaded and adding chargers is kind of a difficult thing because there is no benefit for the flexibility that is monetized, both in terms of real and reactive power. So I think the type of work that Amuro was commenting earlier has a huge potential and it will be very critical in the future. Great, thank you both for that. We had a question come in early on interested in somebody addressing the concept of transactive energy and control. We heard a little bit from Amuro about that concept and what it looks like, how it differs from more traditional structures and resource management and trading, but also wanted to note that this coming Wednesday, at next Wednesday's webcast, we will have Chris Irwin of the Office of Electricity at the DOE discussing transactive energy again. So let us know if the discussion today and upcoming checks that box. And then, otherwise, we've got kind of a nice, I think transition out question. I'm interested in understanding what role universities play in this pipeline, really close to my heart coming from this space when I was back at the University of Minnesota. So interested in the unique role that universities play and then to extend that, what needs do you have to take the work that you're doing from research to commercialization? So you all play a unique role, Ron and Amro. What is that? And then how do we take that research to commercialization move things forward? Yeah, that's a really big question. Ron, do you wanna take a crack at it first? I'm sure. I think that first and foremost, the mission of the university is education. And I think a gigantic role is preparing students who can go and address these challenges in the real world for all the companies and policymaking and so on out there. There's kind of the set of knowledge, the knowledge base they need to have to address these problems in future energy systems is very different from the past. You're not designing transmission lines anymore. You're talking about IoT and communications and distributed algorithms and things like that, but also policy, economics, social sciences. So I think that is kind of the first and I really found a lot of success with the students at Stanford and at other institutions. The second I think point that I think is very critical is the university can serve almost like as a neutral zone where all these different stakeholders can meet and exchange ideas, exchange information, exchange data and work together on problems. It's like a neutral ground because the mission is education and is an art for profit organization. So I think that with that mindset, we have done a lot of the projects that you see here. It came out of conversations with the industry and interactions with very close interactions with them, but with typically with companies that don't talk to each other and then also policy makers and so on. And another example of that is at Stanford, we have hosted lots of interesting data sets where we collaborate in teams so that the owner of the data set is aware of all that's going on, but that has led to really useful tools like wisdom is an example of that. So share data. And I think that the last piece and this is something that I think we need to improve from the universities is really the awareness around these problems. I think in the US today, you see young people much more in tune with the notions and ideas around sustainability, wanting 100% renewable grid, but a deeper understanding of all of the actual issues, the scale of what this transformation means, and then motivating people to go and study this, get involved and work in this industry. I think that's something universities can do certainly a better job of. At Stanford, the Precourt Institute and now Bits and Watts had started a really wonderful program that pays students to do internships at the California Energy Commission, the Public Utility Commission, or in this year, they're even doing it with actual companies like Siemens and so on to kind of understand what's going on in the real world, influence it, but also learn how to then translate this into awareness. So let me build upon what you said, Ram. I think that the universities have a really important role to play as credible, honest brokers. So from my experience in talking to various stakeholders in the state of New Hampshire, one of the first things that they recognize is that we do not have a vested commercial interest in the grid. And so when we are speaking, we're speaking based upon a position of science, based upon a position of what we'll bring about the greatest community benefits, whether it's sustainability or economic benefits or so forth. And that's a really important voice in an energy sector that is oftentimes very much driven by the commercial interests of the particular stakeholder that we're talking about. But in addition to just simply being an honest broker in that sense, there's also a credibility that comes into play. So one of the things that I have again experienced is that credibility amongst technical people, engineers comes from the technical knowledge. And so as I interface with some of my technical colleagues, maybe at utilities or otherwise, they recognize that we understand what power systems is all about or what power flow analysis is or power quality, so forth, so on. But then universities also have an important role in terms of maintaining town-gown relations and being a part of the community and not just an island within the community. And so as we have worked with the town of Hanover, the city of Lebanon, we very much look to see their economies as being very vibrant. So universities have this really essential role to play and because that they can help nucleate a lot of these discussions that are needed to bring about a shared integrated grid. There is a big role to play with regards to thought leadership because once you kind of have a grid architecture, it's very easy to keep it going as it is. But to actually change the fundamental structure of the grid to something new, that really requires some deep reflection that you can't always quite get if you are just chasing after your quarterly goals. And then lastly, I think someone asked about the time scale. We do graduate research and it could be as simple as a two-year time scale for a master's degree or a five-year time scale for a PhD for one or two students to actually produce something that is very viable as a technology, particularly if the technology ends up being some sort of software or IT system. That's been our experience. We've been able to commercialize our enterprise control simulation software that Isano-England is now using for their renewable energy integration studies. So a lot can go, you can go a long way just by having students be aware of what the state of the art and pushing it forward, graduating and then bringing that to market. I agree, Amru. I think this commercialization side is often misunderstood because I think the source of a lot of the interesting ideas is out of thinking about the fundamentals. And out of that, it manifests itself sometimes as software, sometimes as just the people. And somebody asked about what's kind of the challenge to commercialization and having done startups that came out of Stanford in other areas and comparing them to energy, I think there is two really important challenges. One is the energy systems are very critical and today they're extremely reliable. So the time scales at which you need to do a startup are very different. And this actually offers an opportunity for new types of partnerships where companies can identify interesting technologies developed in Stanford or Dartmouth and then co-invest with the university. And there's many different models that are being explored to kind of bridge that time scale. There's also programs like the one in Berkeley, the Cyclotron Road that helps you do that launch. I think the second challenge and this I think it's much more key. As a community in grid, we don't really do an excellent job of keeping what are kind of important practical problems that if you address, there will be a ton of interest. If you go and you're working with medical data or you're thinking about data centers, each one of these communities has it very clear what are some like target problems that can be worked on by companies, researchers and so on. I think every plays a great role there as a mediator of some of this, but if we could start doing that in the long run, it will really be helpful because lots of PhD students, they decide on what they wanna work. And as they look at open problems that are coming from industry, they will certainly give it a lot of thought. Great, for that discussion. We are in a great place today. We're at one minute until the end of our hour together, noting that we do have upcoming webcasts the next two weeks on Wednesday, and we still have questions unanswered. So when we have full discussion and stuff that's left undiscussed, that means we've had a successful day. A couple of additional questions I'll throw back to our presenters and we can follow up online, but do wanna thank everybody for joining us today for taking the time out of your day. And we look forward to ongoing discussion in this space. Thank you again, Rom and Amro. Thank you. Thank you very much for having us and thank you for the audience participation. This is very interesting and the Amro very nice presentation as well. Likewise, Rom. Always a pleasure to work with you, so. Okay. Thank you. Good afternoon and morning, everybody. Good evening.