 Hello folks, welcome to our webinar. We're going to give people a minute or so to join, so just please hang tight and we will get started in a minute. Thanks everybody for joining. We might go ahead and get started as folks are coming in. My name is Anne Jekyll. I am the associate director of New Mexico EPSCORE and co-PI for the New Mexico Smart Grid Center. And we're so pleased today to have Dr. Kevin Tomsevic here to present on the current center that he leads current stands for the Center for Ultra Wide Area Resilient Electric Energy Transmission Networks. Just so you all know, we record all of our webinars and we archive them on our website. So if there's anything you want to revisit today to look at, you can see both the slides and hear Dr. Tomsevic's presentation there. And also a note of webinar housekeeping. Nobody can speak. Your microphones are disabled, but we use the question and answer box in the zoom interface to ask any questions. So please locate that and type in questions as they come up and we will either field them during the presentation or hold them till the end depending on on how it all goes. Next slide. So we wanted to let you know that in May, we have another webinar on the 27th from noon to 1pm. And the presenter will be Christine Fund, who is a nationally recognized expert in research mentoring and the principal investigator of the National Research Mentoring Network Coordination Center. She's also part of a National Academies Committee that recently published the consensus report and online guide, the science of effective mentorship in STEM. She'll be presenting on the current landscape of mentoring and science and summarize that recent National Academies report in addition to sharing best practices and mentoring and available resources to enhance mentoring. She's really prominent nationally in these areas and we're very excited to have her so we hope that you can also join us at the end of May for this webinar. So with that, I will introduce Dr. Kevin Tomsevic who can go ahead and grab control of the screen with his presentation. He is currently the CTI Professor in the Department of Electrical Engineering and Computer Science at the University of Tennessee and Director of Current as I mentioned and current is interesting because it's a National Science Foundation and Department of Energy funded Engineering Research Center. Its scope of work is complementary to the New Mexico Smart Grid Center as well. Perhaps most important on his CV is that he is part of the New Mexico Smart Grid Center external advisory board where he is a very valued member. He received graduate degrees in electrical engineering from the University of Washington and was a professor at Washington State University and the Department of Electrical Engineering and Computer Science for over 15 years before moving to the University of Tennessee. And I triply fellow and very accomplished engineer and we have a full bio with additional details on our website. So we're very much looking forward to his presentation and please go ahead Kevin. Thank you and appreciate everybody joining today and hope everyone's doing well at home and not going to stir crazy. So I want to do I guess two things today the presentation one kind of motivate what we do at current and talk a little bit about the center brag a little bit about what we do. And then most of the time, once I motivate that I want to focus on what we've been doing for for test beds. I think that's we've done got some interesting work in there in terms of how we're going about simulation and thinking about grid modernization. And also there's a bit of advertising in there because we do have this tool and we'd like to get more people using this it's open source. We plan to release one. Well there's parts of some pre beta versions are already available but we think it's pretty powerful tool we would like to to grow our user base. So, first off, let's see why is that not working okay. Let's just a little bit about what ERC is. As Anne mentioned we are semi unique there's two of us who are jointly funded by NSF and Department of Energy. There's another one at Arizona State on solar flotex. That's unusual for NSF in a couple ways one it's long term funding. We're actually in year nine right now. But the other thing is that it's really meant to be a comprehensive approach to research. So that means you want a lot so students working together you're trying to comprehensively solve a problem. And as opposed to a typical NSF project where you might have two or three PIs on it and a couple of students. You know, we try to get all our students working together. It's not unusual to have meetings with five or six faculty and 20 students all trying to coordinate their work. I think that's more similar to, you know, a working environment. And plus I think it's, it's a better way to to make accomplishments and research. In fact, there were the lead at the University of Tennessee but the our partners are Rensselaer, Northeastern Tuskegee. There's an industry program. There's also outreach and education program. I'm not going to talk about those but I'd have to talk to people and they have questions about that at the end. Okay, so first of all what is exactly current and start off by saying there's a couple of main drivers for the changes in the grid. Of course, we're working in integrating high levels of renewables in the in the grid but there's other things that are really changing that are that are helping this along first of all is new measurement and monitoring. Particularly the grid, because of the high voltages and so on it's expensive to monitor. It's not that well measured but and most measurements were not synchronized. But in recent years with the advent of phasor measurement units, PMUs, there's now a wealth of information out there. So there's new types of measurements coming on for those of you not familiar with PMUs it's basically just using the GPS signal as a synchronization source. So you get the state of the grid at a snapshot. But there's all sorts of other sources online you know these days. You know utility that has a hydro system they'll be monitoring wallet water quality in the control center. They've got weather monitors and watch lightning move across the system. You can integrate all that into a view for the operator is one of the things we're interested in and it's necessary or it's going to have high levels of renewables in the grid. The other thing that's changing quite a bit in the grid is the communication. So, more or less you can say the backbone of communications in the grid is based on 1960s 1970s technology. It's a cooling system it's hierarchical. You would get in information from the substations every two to four seconds in the US. And it's not, like I said it's not synchronized. So, if you really want to do some calculations on that you've got a piece these pieces together. And as a result there really isn't much in the way of true wide area control. I mean something where you have to get remote measurements do some processing and then do control from a distant location, which we think is critical for renewing for integrating renewables in the grid. And then the last technology change that's really moving things along is what I've lumped under actuation. And that really refers to power electronics and it could be all sorts of traditional controls and generators as well but the thing that's making changes is the power electronics. So a new modern power electronics which is how most renewables will be interface to the grid. One has a lot more flexibility in terms of control that they can do on the grid. And that's really what the theme of our center is which is the center of this diagram is its control new methods of control. We want to be measurement based wide area and flexible based on a power electronic devices. And a little bit more background here in where we are it's mentioned we're near your nine of our center. And this is just trying to show the progression that we're moving on at this point in time. And we want to be thinking about systems across North America so we think of possibly having a HP DC overlay that cuts across the three interconnects with the ability to deliver more than 50% of energy from renewables, which means you know at times close to 100% say greater than 80% instantaneous power, which is inverter based resources that HP DC overlay to take advantage of diversity of load and timing differences across country, greater control overloads, different demand side management or it's just increased use of EVs and the ability to to trickle charge versus supercharged but responsive to grid needs increased monitoring. And then the last thing that we've started to research more is restoration because this has become a bigger concern for utilities with increased levels of renewables. So here, this is kind of high level. This is sort of philosophy. But we think we get resilience on the grid and scalability with high levels of renewables by being one distributed so there's too many devices for it all to come to the control center. There's going to be some autonomy among the control and then with without large gen central generation to provide the controls, we think demand has got to be an active participant, whether that's aware, whether that's under the control signals or just locally taking action to support the grid. But there's some distributed nature to it. We think there have to be lots of measurements so things can be learning and adaptive, it can be data driven. It's nearly impossible. If you start thinking about millions of devices connected to grid as being your generation sources for you to use a model based approach entirely. Things still be modular or higher colon some sense so that you have a context for the type of controls and by that I mean of, say we're in a high wind day where there's a lot of import from the northwest. So we're not sounding to California, then devices in California that can provide both to support understand that they're in that context they may not need communications but they have to understand the context of operation that day. And then where our philosophy is also that we're sharing resources. So we're not everybody has their own grid on top on the rooftop and they disconnect. And, you know, go off grid and that we think there's lots of advantages to being interconnected that's why we have the grid in the first place. It helps reduce uncertainty and it provides support for failures at different parts of the system. Okay, so that's my, my background and on the on the center research what we do. So what I want to do turn to next is looking at what are our simulation testing needs for this future grid and what's changing about it that we need to think more closely about. So I'm going to, we did two major test beds in our center one is a large scale test bed where we emulate large grids at software based for use me with power systems positive sequence type of models. And then we do a hardware test bed where it's entirely from power electronics inverters that are emulating different devices. So the hardware test bed it's more difficult to scale up to large size, but we think you get more model fidelity that way. And then if there's time I'll talk a little bit about resilience but may not. Okay, so here's some of the things I've already mentioned but just kind of summarizing them how the power grid itself is changing on the power system is changing. You know, moving from large generators to smaller distributed generators rooftop solar or smaller wind farms. Moving from rotating machines that you can take advantage of the inertia to interverter and face which is effectively decoupling the inertia and coupling the physics from the power system. A passive transmission system to actively control TND whether that's through HVDC or fax devices. More and more synchronized sensing an open network. In terms of economics, it's moving from cost driven system to a market driven or some people like say transactive. So this has been going on for some time with the markets at the end of the 90s. But it's being pushed further and further to to the point where market type of decisions are are at the level of controls, or at least that's the purpose of it. And then the mark the electricity markets themselves are moving to to shorter and shorter clearing time so five minute ahead type of decisions. And then recently, due to such things as sandy and extreme weather events. There's been an increasing concern about reliability and resilience. Okay, so how does this challenge the modeling and simulation we do. Well with power electronic interface devices, the very high speed speed if we want to do longer term time simulations then we want to scale up to do large numbers on it's very difficult. There's some very fast time constants in there as long as it's also slower time constant so that is a challenge for software. Also the inverters are completely programmable though you lose some of that familiar dynamic characteristics that you have loads are changing. Recall now what the percentage is but I think most of your modern even home air conditioners are variable speed drive, which creates a new type of challenge. Since they look more and more like constant power devices, although they could be controlled in a way that's helpful to grid. New sorts of protection issues since fault currents from the power electronic interface devices are much, are much lower. The emergency importance of communication so I think of traditional large scale power system simulation, the communication system was not modeled at all but now we're trying to use more of those communications and also do we have the cybersecurity concerns. So, also we have more actively controlled distribution. So that means getting for the power system engineers on here. It's more difficult to separate out transmission and distribution study so historically we just separate them out and just look at transmission as a hard voltage when we're doing distribution and look at distribution is just a load when we're looking at that's getting harder to do is as things are more actively controlled and whether that's for new ideas like micro grids or actively controlled devices such as some some voltage. I'm trying to gotten the name like smart wire these types of things. Okay, and then last thing for the test beds that are, it's an open question and challenges. So what do we really have to do to get reliability and resilience and we're used to empower systems this idea of n minus one security, at least at the transmission level we're just trying to make sure that a any individual element is not going to cause an outage. But when you start thinking about resilience then you start talking about extreme events. You know you're going to have outages with them and you start trying to determine what sort of scenarios are are meaningful for trying to figure out what it'd be a resilient system. Okay, so I mentioned we have two test beds in the center pictured here there's a large scale test bed which is a positive sequence type of model. I'm showing a North American system that we've put together it's highly aggregated but still it's crossed across the country, which we've modified to include lots of wind and solar. And that, and then at the bottom there I'm depicting a hardware test bed is showing a smaller system. But like I said that's a, you can sort of think of it as a power analog computer it's a, it's a three phase system with real voltage and currents, it's not high voltage or high currents like 50 to 75 amps but it's real system so we can use actual communications, actual sensors, meaning CTs and PTs current transformers and potential transformers for getting our measurements and we have protection in there and so forth. Okay, so let me move through the animation here quickly for online it's a little bit easier so the objective for our test beds is, is multifold here you know we weren't we're doing research on on these want to test out new technologies, particularly for new controls. And we also want to be able to test out power electronics and different types control architectures for this system. Since we're talking about a real time system we wanted to include real time communication networks real time control capabilities. Also wanted to be able to be both a source for research and for teaching and testing out equipment. So we went to scenarios that we can do this to reflect not just reliability and secure performance but resilience. So let's start with the, the LTB so again it's, we have a closed real time testing environment. We want to model everything that would go along with that so not just the power system components that we normally model but all our measurement devices, any actuator models, we have. So you can think of this as a cyber physical type of simulation environment real time. We want to be able to model these renewable generation and all the power electronics. On the other hand, we want this to be real time so we have to have aggregated models of these things we can't do, say the switching models for power electronics wants and sort of averaging model. And we want to provide good models for newer types of generation with we get renewables. All right, so with this in mind, we sort of faced a kind of decision here about how, how you want to go about this. And I'd like to think in, you know, in terms of broadly there's two extreme ways we won't go about putting together a large scale test bed like this. One would be a sort of fully integrated approach. You can think of that as a kind of medelica modeling type of approach where you have all your models all in one system. You have this very powerful underlying software medelica that can use different routines to deal with problems like systems and that and use the tools that are there. But this requires the developer to understand all those modeling and it requires a lot of work to get everything together. Now the other stream of depicting on the right is sort of the cold simulation type of approach. I think that's probably best represented by the helix idea that the national labs are doing where these different components that you need are all very decoupled they're developed separately and then they're there when they put together and simulated together. There's kind of a weak link between them. So this has the advantage of you can develop these different things separately say your transmission. Here I'm exact given example grid dying, or your distribution using example grid lab D can be done separately and then you have to worry about the, the interface. So that's good experts can kind of focus on their part of the model. But then the interface here doesn't actually guarantee that there's you're actually getting a true integration here. And typically what they do when the interface is to just throw away a bunch of information that the one of these other modeling devices is giving them. So our approach is a little bit different trying to split the middle and hopefully not ending up with a, you know, the old joke El Camino not much for car and not much of a truck right. But our idea is that we want to have this independence of system but we're going to force some structure on it so we we integrate the models, but we allow our models for some sort of plug and play type of, of architecture so if I can. Think you can see my mouse as I go through here so you can think of our modeling types of things so this is Andy's or in house. So this is a sequence type of simulator that I mentioned at the beginning that is hoping people pick up the phasor sims from our Opel RTV hardware based system. So in our simulation, any of these good dine other simulators. They can be used as plug and play with ours and all they need to do is to follow with streaming protocol that we have in our server. The simulators, they stream out information and skip that information back stream this is handled by our, our dime servers what we can call it, which interface is to different models and renewable models, energy management system models, and measurement models, communication network models. So the, the main thing we're providing here, although we do have these different modules and that is, is that central core thing which allows things to be glued glued together. And we force some structure on this and you can I guess think of this glue is really trying to faithfully reflect a real system in terms of how you would communicate and so on on the system. Okay, and this is we can run in non real time but it is meant to be real real time. Okay, so the advantages for what we're looking for to get to help out researchers and build things is this module architecture. People can kind of focus on what part of the things they want, whether that's, you know, they're working on new control algorithm they don't have to worry about developing new models of the generators. And if they're emphasis on communications and they can work on that part and they've got all the other models for them. And what we have right now was about 1000 in our depending on what simulator using a few thousand bus model but we think we can scale that. So it's, it's got some ability to scale up. A few thousand buses is still pretty decent when you considering your, your, you know, modeling all the communication, all the measurement devices and, and so on. Okay, so this just shows a little bit of the simulation engines I'll try to pick pick things up a little bit here in terms of speed and show some examples so this is our in house one Andes. So we can use some that are more for high performance Q puting like grid dying, or these commercial commercial ones right word now stick in terms of the simulation engine itself. Again, I'm going to, I'm going to skip through some of this because it's a little bit too to detail so I can show some things. But this is our in house. Positive sequence simulator. I'm going to show you types of things we include so we'll, we'll have a simulator to the PM us so our measurements model here we can do that with measurement noise or somebody wants to focus and expand the model then that's, that's quite easy to do in the modular structure we have. We use standard communication protocols on this, such as the open, forget the acronym again. Open, what is it for the standard protocol for for PMU type of measurement PMU concentrator type of measurements communications. We also built a visualization on on top of this so this is partly for educational purposes to show some results of this but it's particularly important for large scale systems since, you know we start dealing with thousands of thousands of buses and tens of thousands of variables in order to actually see what's going on it's, it's challenging. So this is another module that is modular and that we're just grabs streaming data it's completely developed separately from our, our simulators and and whatnot. Okay. So I mentioned what we really provide is that the glue that holds these things together and that's this what we call the dime server. I'm not going to discuss this at all it's just that that's the way we do the streaming of it to tie these different modules together from the from the system. And the other thing is our, our communication simulator that we have for we call it LTV net and it's works on standard communication protocols. So just a little bit on the communication protocol here that again it's modular setup separately. We're doing this in in software for right now, although we can easily substitute in a hardware based type of system for this for for this purposes of what we're doing here on the scale we're doing it seems software is fine there's a few tricks in trying to get all these communication delays and stuff realistic but it's fairly easy and suffer. I see I had one question pop up on implementing the communication network. So we use an open source tool called what is it that tool called I want to say it's something like I believe it's called op net or something like that. Software names always do me but it is the communication is not integrated into the simulators like the question I got here is about how it's integrated with Opal RT. So from the Opal RT the phasor sim system, we would have the data streaming out of that representing the the simulation that would go into our communication network emulation and also go into our measurement and look at the simulation to get the you know the the processing done of that it would then come out of that and then be fed back into the simulator. So it's it's reflecting the actual physical structure of the system. Okay, so this is an example just we have it across the North American system probably one of the weak points about this is just not really knowing much about the actual underlying communication network. We sort of have a follow the the control centers and the different parts of the country. Yeah, and this is just showing that the streaming server that we we have on this this is the name I was trying to think of for the open PDC. So we use standard protocols on this again so that we think it's reflecting close to the the open system. Okay, I'm going to speed things up a little bit here flip through this to show a couple of examples of of what I want. So we have this system with this higher levels of renewables it's interconnecting the three in the the Western US, Texas and the Eastern interconnect with this HVDC overlay that's represented by the green lines you see here. And if you can see on your screens got these some wind turbines here some solar down here and so on. And we've got roughly 50% wind in this. Okay, and so what I what I want to show here in a couple of examples is the integrated integrated approach we have here, but in the types of things we can do on this, but really the whole point of this is that it's that you've got all these things that you've got to consider for the system, the measurements, the communications, the underlying physical devices, power electronics devices integrated in one package and this forces us as researchers and as well as our students of course to to really think of these things comprehensively. So I'm going to do not going to talk too much about the techniques that are underlying this but just going to show off the the tool. So this is going to be a case where I'm using a modulation of the renewable resources. I think this one or maybe it's the HVDC. Yeah. To yeah it's really I'm sorry it's with the wind turbine so modifying the controls on the wind turbines to emulate inertia that you would normally get with a synchronous machine. And let's just run it here and hopefully it comes out fine over zoom. So there's a disturbance down in in LA and what you're seeing here and the and then animation is the variation in frequency so it's a little bit tough to probably tell over this presentation but you shall run it one more time. But in the uncontrolled case you can see there's a there's bigger frequency swings and it takes some time to to damp out. So it's it's it's really showing the advantage of using the power electronic controls that the thing you notice is there's a poorly damp mode over here on the on the right side. And the reason that gets excited from the disturbance in the West is because we've interconnected the the West and the East through the safety DC overlay. So one more time. Here's the disturbance. You see it swings here and it's still swinging on under controlled case. And here's my local mode over here. Okay, so like I said this has got all the different components of the modeling in it. Let's go to the next one. Okay, this one is, I'm just going to focus on the Western US this is work done by a student mind assigner alpha. It's now quanta where the idea was we'll we've replaced a lot of the traditional generators in in the WEC with wind farms and now because there's still some poorly damp modes out West, there isn't the actual system. We need to provide some some damping using by modulating the wind turbines. Again, not emphasizing the the techniques underlying that it's distributed control technique, but to show the the value of this tool. Okay, so again, there's going to be disturbance here and then from the uncontrolled on left side controlled on the right side. And this is a, I believe an East West mode in the southern part of the system. So again, let's see, we get the trip. This is showing the actual voltage you can see now things are starting to settle out on that and then you see this poorly damped mode that's still going on between the Rocky Mountains and into the into the southwest. Okay. The other thing about this distributed control idea here that we're testing out is that it's very robust with respect to failures. That was the work of a sign. Okay, let's I think I've got one more. Example. This one is just again showing off capabilities of the of the of the platform. It's demonstrating a cyber attack. It's false data detection attack and makes leads the operator to believe for the control to believe on close up control that there was a a major disturbance and it leaves you a mistaken protection action of separating the system so you'll see two island form in the WEC and you see this there's the false data looks like things are are separating this causes the frequency deviation. It thinks there's a major event happening and then protection acts to separate it so now you see these two islanded systems. Okay. All right, so to kind of keep things on reasonable schedule here. Let me now move into the other part of our test beds and that is how do I get me to answer that one question live. I don't know if everybody can see these questions. So let me repeat it is the false data injected in transit or by compliment devices. This is by compromising the in this case the communication system so in transit. So as I believe it's a man in the middle attack. Okay, so the other test bed is this hardware test bed. And so we now are going to take highly reduced models of each of the three interconnections but this is an analog system so it's three phase it's real devices kind of have the same goal that we have with the large scale test bed and that we want to integrate all these different components together and that forces you know new things we try out to be to be faithful. The way this works is we have a DC bus we see the power grid here this is the actual you know supply coming into our building. So we have a DC bus and that then we have different emulators for generator for loads. Excuse me I'm getting a phone call here I don't know how to shut that off. Okay, the so the that we have our emulation the power system and then the what we only have to supply is the losses right so we circulate the the power in here. This just shows a visualization in our control room we try to make it look like a real large system so we've got these different stations for we have four areas represented in this case and when we run it you're actually looking at as if it was three four different area controls although we do have a central controller for doing it. Okay, so the advantage of doing this all in hardware is that we can get a wide range of time scales in one system so we don't have to worry about simulation type problems this is true parallel processing because it's actual hardware right so you can't ever in software do true parallel things it's it's just you there's got to be timing things but this is real devices in parallel so we can do that and we can get the full range of time scale so it's true multi scale. It's got the real time communication and protocols and protection and that for that. Our the architecture of our power electronics is is essentially the same as it would be in a in a real system so that's very faithful reproduction. And we can also interface this to things like an RTDS or other time real time simulators. Okay, this is just kind of showing the different types of things we can do different types of loads and generators we can emulate wind solar. Let me just point out there's a lot of color coding here have to do with whether it's a inverters voltage type or current type. I think the one of the things that's interesting is we can actually emulate transmission lines so we can reconfigure this system. So it is all in a hardware but we can reconfigure this different systems of different transmission line lengths and different treatments like characteristics. We don't do that for all the lines a lot of lines are emulated by inductors but we can mix and match our inductors to get different systems on it. So I'm going to I see a couple more questions coming in but I'm going to try to I'll get to those and let me get through things here so I can get through the questions. And again, we want this to be a fully integrated environment for for testing things so that this is showing all the controls is a very busy slide but this is what's, you know, a real power grid would have you have at regional level you have things like state estimation and dispatch. You have your AGC automatic green control for maintaining frequency. At the local plant level you have protection and other types of controls you'd have stabilizers a troop. And then at the systems level central functions level you might have some other coordination type of functions. So when we want to test out new controls in that we want all this functionality to be there and that's our big emphasis here on on integration. You know you think about typically what what people do research might have a simple model and try out new control but they're not thinking about all the other parts of the system that are here and here everything is there and it's running all the time or while it's supposed to be running all the time may often students will do things like shut off certain the controls because it's harder to see what the impact of their control is but we intentionally want that to be there because we intentionally went to see to really test out new ideas and so on so forth. Just to show one example of of something that you can can do with this that you would never be able to do on a large scale simulation is that we're able to discover when there's residents and stable harmonics that are building up between inverters so this can happen when you've got weak links and you've got multiple inverters tied together and you can develop these residents and it's usually fairly easy to to to mitigate by adjusting control parameters but without this type of emulation you wouldn't be able to discover this in your in your control design right this is something you could never do in your traditional power system simulation or we couldn't do in the LTB because we don't have the detail of power electronic devices model there and here we have the actual power electronic devices show another one that we can just because it's unusual that we can do in power electronics is this idea of emulating transmission lines so what we do is we do a back to back DC and we use a Bergeron model to do this so we can effectively vary what this reactance is here I'm showing it on a two area can do a system but we can vary the reactance is here to make you know look like a different system different line lengths or we can even do short circuits or or open line faults for this thing so again something to very difficult to do on a on a positive sequence type of large system model another thing we can do is these models for voltage source converters I'm not sure how familiar everybody is with with different inverter types but voltage source inverters are would allow let's go to multi terminal systems and then we'd get the robustness and reliability you get out of network systems but they're very hard to control so but we have those implemented in a system in fact we have those running on the HP DC on our HP DC overlay on the system now okay I think in the there's only one actual operational multi terminal HP DC in practice which is in China just the terminal guy okay and then lastly I think this is going to be last I'll get one more thing at this is the as we we do implement the type of protection here we have real protection this because it's real current but we also do system type of protection schemes which you implement system again because everything is is real components and we can do these things in in parallel without installing stay in real time so if you're familiar with you know what you do in large scale simulations then normally the protection isn't modeled at all okay now we're not doing the detailed protection here for what the actual signals kind of waveform we look like but we're getting the types of protections you would get in for things like under voltage and under frequency and other generator types of protection okay so lastly then and I mentioned this earlier is that we can interface this to other systems we have to go through amplifiers to from the something like an RTDS but it allows us to expand this HDB and we've we've done that and this is a showing it this is showing our our control our room control room on the right here with all the hardware in it a little bit hard to see what's see what's in there but you can see a little bit here what the you know there's real protection devices in there there's switches for your configuration and power inverters over here and the system I'm showing from North American overlay here everything except for blue dots is implemented in the hardware testbed so we've got this a highly aggregated model of the Western US system we have this HP DC overlay and green here which is in the hardware testbed and then we have air cut in red here which is also in the HP DC and then this also hydro cold backup here through that DC link which is the way it is in practice we have that in the HDB and then all the blue things here we have interface to an on an RTDS so we can run scenarios across across the system okay I do want to leave at least 10 minutes for for questions and getting back to those so this all and on on for the HDB is is it is reconfigurable and we it's essentially like a microgrid maybe different voltage levels so we can reconfigure it and emulate since there this is a project we have with the Chattanooga utility ETB this is a microgrid out at the Chattanooga airport emulated on that HDB and allowed us to do much more involved testing for the controllers before going out and implementing it this is now in that stage of field test okay and I'm going to skip the things about resilience although they'll be in the slides and just acknowledge our supporters these are companies or members of the center that provide support and of course NSF and DOE for funding research so with that I'm going to wrap it up and let me back up so some of the questions and catch up on those I think the let's see the back up to the question let's see are these in which order they kind of come in okay I'll take them in the order that I see them from time so may I know that fall state injection attack on which parameters in the system in that one it was on PMU measurements so the interesting thing there actually we've done other work on that which is that if you really a number of people have looked at trying to influence the state estimation in that if you really look at a good state estimation implementation false data injection attacks are very very difficult to get past the state estimator and not without some sort of sophisticated insider attack so we have some work on that but in that particular example we weren't running the state estimator from Jamie can we use message authentication codes and also students choose to address this problem can it be assumed the devices are capable I think you're I think this question is referring to make can we harden our measurement devices which I think that's that's clearly true the other way is to do it at the system level because when you have false data injection attacks if you then do any sort of processing for things like the state estimator and have some modeling of what the real system performance is it gets very difficult to fool the it's very difficult for a false data check to check not to be detected I'm not a cyber security person but I guess I usually like to think in terms of layers of protection right so there's at the you know at the encryption level at the different ways of hardening that but there's also the system level the physics of the system that we can do it hi Kevin this is Jay thanks for answering the presentation was really really nice to learn a lot in fact we learned so much that we are now wondering why we haven't been collaborating with you guys so this is probably a great place for organic collaboration so one thing is as a cyber security person that I've been to many of these conferences and also read papers and stuff so one of the things that we do on the cyber side on the internet is to prevent this notion of false data injection you use certificates and what you do is let's say I want to communicate as a PMU to a PDC or a super PDC let's say I have a shared key between the two of us I create a hash of the message and then sign it with a shared key at which point anybody who is in the transit path between me and the PDC cannot change the data they change it then the the Macs are not going to match when the other side decryption so the question that I had was I don't hear that very often in the you know power system area or even people doing a surgeon is it part is it because the devices are not capable or are we looking at backward compatibility or because you know your Raspberry Pi's can do these things easily I understand your question quickly I think it's mostly because there's not you know these utilities think that all this stuff is well all this stuff is on private networks and utilities think they're pretty secure I see so that I think that's by that definition false data injection shouldn't happen um that's that's true it I mean I think it is I think utilities are much less so you see much less coming out of the utilities on on that false data injection attack than you do out of researchers who are worried about it however I would say this I think that you get you know anytime you start having these you know the other thing about the PMU measurements is nobody's using them for control right now so it's hard to get the utility excited about a false data injection attack up because they're not using control they're just using them for data logging but also there are protection PMUs I don't know if people are using protection PMUs the utility side there's a lot of talk but I guess there is there are some there is some protection I don't know if it's wide area though so I don't know if it gets out of the the sub substation but I think right you know we're researchers we want to look forward and and sooner or later things will be wide area and I think there will be points that where they you know the communication could be exposed whether it's at the the PDC I don't think the open PDC has any security protocol associated with it right that's interesting I mean I have to profess ignorance I actually looked at the standard or the implementation but I always find it interesting that you know when I go so I was recently at ISGT and I asked a couple of utilities this I was like you know you guys are talking about false data injection first of all if you're as you rightly pointed or something like yeah well you know people are doing research but we are not that worried about it I'm like why aren't you worried it's like well you know we have a private network and everything I'm like so are you guys air gap they're like no we are technically not air gap I'm like well then you know it doesn't matter how private you are your networks are potentially open for attacks that is one I mean again as a cybersecurity person I always take the pessimistic approach as you can see from my questions but the second thing was you know even the incorporation of these you know message authentication codes and signatures in the communication domain people have been studying this for 20, 25 years and there are very very very efficient ways of doing it and you can maintain the transmission rates as you have right now with very little overhead so that should be an interesting exploration even yeah I think I'm not a cyber person so I'm not familiar with all of those techniques I think the I think one of the other you know the main reason why the utility people aren't concerned about it like I said is they're not really using the PMU stuff for anything in its real time now the one thing so for real time controls the one thing that could be an issue might be since they do use it after the fact for clearing the market and what the actual sales are so then you could false data there which could impact the you got to pass the state estimator which is not easy either you got to pass that you could affect the actual settlement so there could be a money maker for somebody so I've seen there's a couple papers out on that but you know over time I think whether it's protection or other closed loop controls utilities are going to have to think more carefully about it right so we have just a couple more minutes I want to make sure we get to these other questions if we have time I'm done thank you we'll move on to Anjouz go ahead Kevin yeah LTB net is it is a communications network similar it's built off of and like there's too many acronyms around for different software packages built off that open open software it's I think I want to say it's is it mini net or op net but it's an open source software based system okay let's see the next one energy source in theory yes all that I mean we can have unbalanced conditions however right now we we're not running a ground so we don't have the third we don't have that the actual return we could do it but it's a little bit complicated I'm probably going to add that functionality because we want to do some microgrid things but yes it's where we're just running three phases we don't have that fourth wire there's certain unbalanced things we can't we can't do we can't do anything zero sequence obviously let's see I'll give see one more here as you work with many different entities and organizations what strategies you implement so like coordination let me let me say this and this is one reason why I talked mainly today about the test beds I don't know if this is really answering your question but that's the way we collaborate it's because everybody has to in the center they develop some new device of new measurement technology they're supposed to get working on one of our test beds one of two or both the test beds and that forces people to to collaborate across research areas you have to specify things it's it's sometimes a lot of work but it's really forces the research to be more closely integrated if your questions more about the organizations that a little bit similar answer and that we try to do that even with the other universities that are at different campuses as as we want we have students visit from different schools come here and for our hardware test bed and and implement their their strategies so I don't know that answer the question but we do use the test beds for collaborating across research groups that's great thank you so much Kevin for this presentation which I think has sparked a lot of thought and ideas for our own research crew so echo Natsuraj's comment to thank you for this presentation remind everyone that our next webinars on May 27 from noon to one and current also has a website and I think some of their own talks that they broadcast so you could visit that website to learn more about the project so thank you much very much everybody for your time today and attendance at this webinar thank you and bye