 So, today I'm not going to talk about Stanford or Bits and Watts and I'm not going to talk about Gizmo because I know that other people aren't going to do that during this week. So I'm going to, and I'm not going to talk about grid innovations because I think you've had, you will be hearing a lot of overview of vision and things like that. What I want to talk about is a specific area of grid innovation and what we do here at Stanford and Slack and I want to expose you to a lot of things that we're doing so that you can think about when you start your own research, the types of thing that you can, you can do with us, with the Stanford faculty and the Slack scientists. All right, so just one slide on Bits and Watts and Gizmo. Gizmo is a group of about 15 scientists and students and postdocs up on the hill. You're going to visit the Gizmo lab if we have electricity on Friday because we're going through a renovation. And then there's another lab here at Stanford called Bits and Watts. It has been under construction over the last year. We're expecting to open it in November. So there's a hands-on, both places have really good hands-on facilities for you to do your research. Let me talk a little bit about modeling and data-driven modeling. We know as engineers you probably have been exposed to modeling using, describing really physical phenomena using math. That's what we typically do in engineering. But there are other really complex phenomena that we can't necessarily describe with math. And what the interesting thing is when these two phenomena, the behavioral phenomena and the physical phenomena come together, it becomes even more complex to actually model. And we know that data-driven modeling has been very successful in these areas. We use tools, we use technologies that use data-driven modeling every day. And there has been significant advances in certain areas, but not necessarily in electricity. But there are a lot of enabling technologies. And today, this is a project sunroof from Google. What they do is you enter your address and it calculates what kind of solar PV potential your roof has, calculates also the shadows and things like that and gives you more or less an equivalent generation for your home. And also, they sell you some ads at the bottom. So if you want to buy your solar cells from any of the providers below, you can go and call them up and get prices. Imagine using this technology, not for selling rooftop solar, but trying to figure out what the potential of rooftop solar in a distribution area is, and how that generation can impact your distribution network. Or if there's a solar eclipse, what that impact is going to be on that local distribution network. So there are a lot of tools that we can sort of think about bringing to the distribution and transmission system planning and operations. There's also a lot of things that are happening in the machine learning applications on the power system side. Apology detection tends to be a very difficult task for a lot of utilities. Because they don't have recent data, it's very difficult for them to figure out which meters are hooked up to which feeders using which switches. So we can actually use machine learning on power systems to actually extract the topology of a distribution network as well. And this is a really great time to sort of take machine learning or other data driven applications and marry them with power systems. Power systems is a very traditional field that comes with really heavy electrical engineering folks. And there is a gap between those hardware driven electrical engineering folks and this new technology, and this is what we're trying to bridge here. There's growing awareness to digitalization. Any kind of a company that you go and talk with have a digitalized or digitalization group within their company. There's a lot of data availability, and I'll talk a little bit about that. There's plenty of cheap storage now, widely available platforms like Google Cloud, AWS, Microsoft Azure. There's plenty of advances in data science and other fields that we can think about applying into the power systems. And then also advances in machine learning and artificial intelligence that we haven't really scratched the surface in the power systems world. So let me talk a little bit about the data. There is plenty of data and there isn't. The reason for that is that some of this data is proprietary and some of it is kept proprietary because of the privacy and security concerns. So that one area that we need to really focus is how to overcome these security and privacy issues. But on the power system, there are many sync phasers on the transmission system. These are sensors that do sampling of the electricity system every 120 samples per second. There are micro sync phasers on the distribution system and they are also working as a sync phasers. And in the western interconnect area, we have about 300 sync phasers, about 75 of those are in California. Micro sync phasers are new technologies that are just being put on the distribution network. There are plenty of line sensors that provide data back into utility SCADA systems. In buildings and distributed energy resources, we've had interval meters in California for buildings over 200 KW and above for many, many years since 2003, I believe. We've had smart meters starting 2009, 2010 timeframe. We have a lot of trend logs from buildings that we are not utilizing today, but they're there for us to be using. Then there's the mobility area where there's a lot of charging session information from a lot of charging vendors. There's a lot of driving patterns that the cars are right now compiling. And some of the OEMs already have information on that. So what we're not doing is we're not doing pure mobility, pure power systems and pure buildings into our work. What we're doing is really looking at the intersection, building to grid transactions, vehicle to building transactions, vehicle to grid sort of concepts. These concepts include technology development, market transformation and understanding the regulatory framework. And sometimes really changing that framework as well. So it's a very sophisticated area because if your technology is good, you're not going to make an impact if you don't understand its value in the marketplace or the regulatory context that it's going to be working in. All right, so I want to talk a little bit about the, I guess, five projects that we have currently either starting or wrapping up, just to give you an overview of what we're doing in terms of machine learning in power systems. So, Vism, Bader, PowerNet, Script and Grip. We like acronyms, you can tell. And we tried to come up with some catchy acronyms. They're all acronyms for a reason, I'll walk you through them. So Vism stands for Visualization and Insight System for Demand Operations and Management. This was an DOERP funded project where Dr. Ramarjagopal received about one million customer data only used about, I believe, 100 to 200,000 customer data within Pijini's territory, which is this territory. And build a way for the utility to design and evaluate their programs. These programs are energy efficiency programs, demand response programs. As a utility, you want to approach your customers that have the most potential. So with a 20% sort of approach, you get 80% of the yield, right? These kinds of tools allow the utility to understand their customers, understand their consumption patterns, cluster different customer groups in different locations, and really provide services that will provide the best yield. The tool is available in the being used by utilities right now in California. It has data cleansing and management capability, demand future extraction, targeting segmentations. These are very valuable to the utilities, obviously. Response models, how fast, how steady the response is being provided. Measurement verification capabilities for energy efficiency programs and forecasting and pricing capabilities. And there are lots of companies who have been involved over the life of this project and continue to take this concept from Stanford up into their products and add additional futures to it. The other project is called VADR, Visualization Analytics of Distributed Energy Resources. Here, the idea is that, what if we have all this data? Can we actually start using all this data to better plan and operate the electricity grid? Now, VADR doesn't go as far as operating the electricity grid, it's all about planning, it's a planning tool. But the idea is that we need an infrastructure to handle all those data that's coming from disparate resources. Once we have the infrastructure, we need to develop a user and developer group that can easily use this data and visualization insights extracted from the data and the developer group that can potentially continue to maintain these analytics and grow the capabilities. And in the end, we want to look at what if and what now, what is happening right now on the grid, create some situational awareness. And then what if, what if things change over time? What would be the impact on the electricity grid? And how do we plan for a future when things change? I won't go into the details, but just so you know, the data that we are receiving is coming into different formats, different time series stamps, they're missing data, get really messy. One of the big problems is, how do you injustice data and how do you actually clean it and how do you use it for different analytics? And different analytics have different uses or need different data sets. So how do you actually condition the data to fit into the analytics that you want to use? So those are important pieces that we're adding. Just one example of the analytics, this is a situational awareness capability, but also has an impact on the utility bottom line. Monitoring each switch is very expensive, but the utilities would love to monitor each switch so that they know at any given time what the switch condition is. What we're doing is, we are taking a system like this. These are all switches, the circles are switches. Instead of monitoring each switch, the orange circles around the black ones are sensors, so there are open switches and closed switches and circles are sensors. We're saying instead of monitoring each switch, let's do a subset. In this case, we're monitoring four out of ten. And see what the limits of reconfiguring the network based on the switch configuration. So we have been very successful with this kind of analytics. And we can recreate the system fairly accurately, even if there are delays in the system. And there are papers that show this as well. The other project, probably Dr. Ram Rajagopal is going to talk to you about this project. But we have an RPE between Stanford and Slack, where we're looking at developing three types of technologies. A smart dim fuse at the home that operates at these time scales. A home hub, a gateway device in the home that communicates with the smart dimming fuse and all the appliances. And then a cloud coordinator somewhere in the cloud that communicates with the variety of home hubs. So when we get a signal from an independent system operator saying move your resources this way or that way. It comes to the cloud coordinator. The cloud coordinator sends instructions to the home hub. And then the home hub sends instructions to all of the connected devices. So this is a project, a three year project. We're almost done with a three quarters of the project. So it's fairly new and there's a lot of potential to get involved. Script, again another acronym, smart charging infrastructure planning tool. Here the idea is that should we actually plan for a charging infrastructure that is just enough to maintain the charging of the cars? Or should we actually build out this infrastructure a little bit more so that there's some flexibility and we can utilize that flexibility for the grid transactions? So this project is looking at that. It's funded by the California Energy Commission. While we're going to do a lot of forecasting and charging, creating user interfaces using a platform, doing a demonstration. We're also looking at benefits and costs analysis of this kind of trade off. This one doesn't have an acronym, so we failed. But I'm going to work on this one. But here the idea is that we have two locations that we want to show the impact of smart charging on the distribution transformers. So we know that charging impacts the distribution network and the distribution transformer. Actually we don't know how much or so what we want to do is actually get a baseline first and then once we get a baseline start implementing smart charging algorithms to show their impact on the, and reduce their impact on the transmission network. So two locations, we have Stanford and we have Google campus in mountain view with 900 AVSEs that we're going to be working with. And Google's going to do a one month test with us, which we're really excited about. Because we don't get to do these large scale tests very often with companies like Google. My last slide is GRIP. GRIP stands for Grid Resilience and Intelligence Platform. Here I'm not going to go into detail because I don't have a lot of time actually, but we're going to do anticipation analytics using data from, again, disparate resources. But here the idea is to build predictive analytics and think about, what are the vulnerabilities on the grid? What are the vulnerabilities if it's a heat wave event, if it's a flood event, if it's another extreme weather event or if it's cyber security event? And then once the event happens, can we absorb this and ride through the event with controlling devices in the field? And then once the event happens and the lights are out, how fast can we recover from those events? So we're going to be building a platform with Google X, look, demonstrate the absorption with Vermont. And then recover from events with a utility in the midst, small utility in the Midwest using the distributed energy resources. My last slide actually, final works. There's lots of exciting work that's going on. And I would encourage you to get in month. Come to the smart grid seminars. We will have them every Thursday. And then we have Thursday happy hours, so come to the Thursday happy hours and be aware of the funding opportunities. Bits and Watts has seed grants that you can talk with your professors about. And then fellowships, so talk to your professors, tap into those resources. If you want to be a part of the Bits and Watts mailing list, please send an email to John. If you want to be a part of the Gizmo student list, send an email to me, we'd be happy to do this. After every talk that I give, I get about 20 students who come and I have the first year I had meetings with each 20 student. This year I'm not doing that. Go start your semester, think about the workload you have. I know these are all really exciting and you want to do something right away. But take a, go through the first six weeks of your classes and then let's talk and I'd be happy to. I'd be happy to work with you and get you more involved in any of the projects that we have. Great, thanks, sir. Thank you. We've got a couple of minutes for questions. And so rather than waiting six months or six weeks, you can ask a few questions. Yes, you can ask as many questions as you want. I'd be happy to answer them on email two, through email two. But as I said, go ahead. There are a lot of small micro grids that people, people have named them nano grids even. So everyone's definition of micro grids is different. You can have a small system where you have a solar charger in your backpack and a little battery in your backpack that charges your phone. That's a micro grid. Anything that's disconnected from the grid is called, can be islanded from the grid is called a micro grid, right? So you can create these micro grids as small as a backpack that charges your phone as big as a huge industrial micro grid. But there are lots of examples of it and everyone's definition changes. The only requirement is that you're eager to learn and work hard. We'll teach you everything else. Can you say that again, I'm sorry? Yes, so distributed energy resources like the solar PV is a huge part of the destructive part of the system, right? People buy and put it on their rooftops and utilities have no idea who installs them, how much it is in terms of the generation capacity. It's very difficult for them to incorporate them into their planning. So some of the things that we do is, for example, solar disaggregation. Which customers have how much solar generation using data analytics? Well, some of these switches are, well, it's not the switches, but it's the switch and the sensor combination. Putting the sensors is expensive. And if the switch doesn't come with the sensor, which is the old switches, then you have to actually disrupt the system and put the sensor. Because these switches are not like your CTs and PTs. You have to actually cut the power to be able to hook them up. So that adds to the cost, labor and downtime. Well, the biggest questions in my mind on demand response are not technical questions. I think the value of demand response going forward is a big question. And that's why a lot of companies are not investing a lot in demand response. I think the technology in terms of automation, in terms of integration, are there. One of the biggest barriers is that demand response hasn't been operationalized in the distribution system. So that is the biggest problem. When you don't operationalize it, you don't know the value, you don't know how to use it, you don't trust those resources. That's the biggest barrier. Yeah, so visitum is something that the utilities are implementing. Vader is something we're working with the Southern California Edison on, and they're planning to implement it. That's why they were really excited about the GRIP project, because it allows them to embed some of these capabilities. I think we're very new. Vader's only about a year and a half old. GRIP hasn't started. So by the time we're done with our projects, we think that there's going to be a lot of embedded capabilities. NRACA, which is one of our biggest supporters in the GRIP project, has about 900 utilities. So they provide these kinds of services to them, and I think the impact is going to be huge. Please join me in thanking Solar once again. Thank you.