 Good afternoon, everyone. I'm Selena Keneally, the Associate Director for New Mexico Upscore. And together with Brittany Vanderwerf, we're hosting today's smart grid seminar. Brittany's running the show behind the scenes and she'll be moderating the questions for us. Before we get started, I just want to do a little bit of housekeeping. We are recording this seminar and we'll make it available on our website probably sometime next week, so please know that that is available to you. And also, we'll have a loud time for question and answer for all of our speakers today, so please use the question and answer box here on your Zoom screen and you're welcome to ask your question at any time and then Brittany will moderate those and ask the questions of the speakers in between times. I'd like to welcome you today to the smart grid seminar which is part of the New Mexico research symposium which we host every year in collaboration with our partner at the New Mexico Academy of Science. Of course, this year we're an entirely virtual conference, and we've had events taking place across this entire week. We started November 9th with a wonderful keynote address given by Dr. Betty Korber, which will also be available on our website, and it was certainly timely because she spoke about virus research and the things that are happening at Lannell in support of the cure for COVID-19 as well as vaccines. The poster session opened on Tuesday and there was an opportunity for people to view the posters and vote for their favorite. I believe that voting is now closed, but the posters are still available. And then today you can see the smart grid seminar which is featuring students from the New Mexico Smart Grid Center at New Mexico State and the University of New Mexico. And I invite you to join us tomorrow, Friday at three o'clock for the award ceremony where we will be recognizing outstanding service to science, outstanding science teachers, and as well as the winners of the poster competition. If you'd like to know more about any of these events or access the materials, please visit our website I put a cut lead there because the URL is a bit long and that website is will be open not only through this week for the research symposium but probably through the rest of the year so you have opportunities to to view and participate in those things even if you weren't able to do that earlier in the week. So it's just a couple of minutes to give you some context about the New Mexico Smart Grid Center. It is a project administered by New Mexico upscore, and the purpose of upscore is to support the research capacity of our entire state. The Smart Grid Center is an NSF funded award it's $20 million from the federal government plus $4 million from New Mexico cost share in support of pursuing research and workforce training for next generation electric power production and delivery. We've just started year three, and the Smart Grid Center itself actually is more of a virtual center and let me show you who's involved in our Smart Grid Center. So we have the three research universities in our state, starting with New Mexico State in the south, New Mexico Tech in the middle and then you and I'm here where New Mexico upscore offices are located. We also collaborate with a community college Santa Fe Community College, both our national laboratories Los Alamos and Sandia Explorer is our outreach museum partner, and the microgrid systems laboratory is a nonprofit as part of this. And you'll notice that we also have nine industry partners which are featured at the bottom of the slide. The Smart Grid Center has a number of research thrusts and you guys will hear about four of them today, and we have, I think about 140 people who are working on this effort, both on the research side as well as the education workforce development and outreach side of things. So without further ado, I'd like to introduce our speakers today. We're going to hear first from Shuba Pati from New Mexico State, then Jesse. Oh, I knew I was going to mess this up. I wrote it down. Coach Marski from UNM, Ali Garasi from UNM, and then Anju James from New Mexico State. So one of our students will have an opportunity to do a brief presentation in about 15 minutes. We'll pause for questions, and then we'll move on to the next student. If we have time at the end, we'd be happy to entertain some more questions. And so without further ado, I'd like to introduce Shuba and let you know that she works with Dr. Satish Rana Day at New Mexico State and Shuba if you'd like to unmute turn on your camera and I'll let you take control of the screen. Thank you. Hi everyone. I'm Shuba. And today I'm going to talk about the resiliency announcement of the smart grid, considering the time varying priority of dynamic load. What happens to the grid in case of a natural disaster, like hurricane, or in case of a cyber physical attack, a part of the grid get damaged. Some of the lines get cut, and some of the generator might fail. All of these things normally leads to shutdown of the power system. For example, in case of a hurricane Sandy, we had $50 billion loss. So how do we define the resiliency? The ability of the power system to recover either completely or partially from adversity is defined as resiliency. Now the resiliency is defined by a term called adaptability. So adaptability in biology is the ability of an organism to respond and survive in case of environmental distress. Similar phenomenon is expected to happen in case of our system. Now, the next thing is how the resiliency is different from the reliability. So in case of resiliency, the priority is to quickly recover through active management of the grid. Whereas the goal of the reliability is to primarily maintain the continuity of the service. And how do we do that? We do that by making the infrastructure redundant. One more key difference between the resiliency and the reliability is the resilience, the objective in case of resiliency is to maximize the throughput. Whereas the objective in case of reliability is to minimize the cost. One more thing is different between the resiliency and reliability is the resiliency is in case of resiliency, it is time sensitive. Whereas in reliability, we focus on the continuity of the service. In case of resiliency, we focus on repair emergency actions. Whereas in case of reliability, we focus on safety and over anticipation. Now, in case of resiliency, we have two parts. First is the resiliency oriented design. And the next is the resiliency oriented operation. So basically in case of the resiliency oriented design, we focus on basically the infrastructure enhancement, like the strengthening of the vulnerable component, increasing adequacy of the power supply, increasing topological flexibility. Also, in case of resiliency oriented design, we focus on some of the specific design action where we can enhance the resiliency, like upgrading the flowhole classes, adding the transfer skies, installing backup generator, or adding sectionalizer. So for instance, we have taken IEEE 24 bus reliability test system in our experimental setup to demonstrate the resiliency study. As you can see in this picture, we have taken IEEE 24 bus system and we have taken two cases. You can see the case one is considered one part for the tornado or hurricane or any kind of natural disaster. And case two is another part for another part for a natural disaster. And we have taken two critical loads at bus 19 and bus 20. Now, in case of resiliency, so we need to define the objective function. So here the objective function is to minimize the throughput or which is same as minimize the, maximize the throughput and minimize the mismatch between the total generation and the total demand. This is like an optimal power flow problem. In this objective function, we have constraint on generation as well as line capacity. Along with that, we have taken additional constraints so that we can always supply the power to the critical load at every point of time. Now, when we simulate the case one, so you can see there are some lines that are greens and some lines that are red. So the lines that are red, they are the lines that are getting overloaded and the lines that are green, they are getting underloaded. When I say overloaded, I mean the load in the line increased by 30%. Similarly, we take the case two and or the part two of the tornado or natural disaster and we see the lines that are red are getting overloaded and the lines that are green are getting underloaded. So the study of multiple resiliency test cases on a simulated model of the power system can infer about the changes required to improve the resiliency of the grid. Like in our case, identify the lines for which that line capacity needs to be increased or decreased. So we can similarly simulate other disturbances scenario and also design consideration can be made to enhance the resiliency. Now our focus is on resiliency oriented operation. In resiliency oriented operation, basically our priority is mainly optimal scheduling. When I say optimal scheduling means the improving the resiliency by local supply of load or by curtailment reduction. So which basically deals with unit commitment, energy storage schedule, adjustable load schedule or energy management system. Now, in this case context, one important consideration is the time varying criticality of the load. So there are operational uncertainty in the load. Certain load doesn't maintain same level of criticality at all the point of time. For example, the power supply to a subway train network is critical for a certain time after a storm or a hurricane or a natural disaster which would allow the train to settle either at destination or at nearest station. However, once the train are parked, the operation could be halted temporarily to ensure the safety and ventilation. During this time, the subway train network may no longer act as critical load to the grid. Now, in addition to the time varying criticality of the load, we also know that DER like wind generators are affected by disasters like hurricanes. In case of hurricane tornado, the wind power generation is a function of time. So in order to model that, we added this constant. K critical, as you can see here, K critical is the set consisting of the indices of all the load with time varying criticality. Where lambda represents the fraction of the K critical load that needs to be supplied for a given disaster scenario as a function of time. Then we also assume that the load vary randomly with respect to time. Here, the KDR is the set consisting of the index of each DR and mu is the fraction of the power generated by the KDR for scenario S at time T. Now, the power demand of load is assumed to become comprising of two random variables. P and one for RVIP and one RVQ, one for active power and one for reactive power. And hourly normalized load profile as you can see here, and we have taken mu and mean and standard deviation is 0.02. So, RVIP and RVQ denote the random changes in the demand of real and reactive power respectively. And here we have taken the normal distribution. Now, we refer to the resiliency curve to determine the frisality of the grid or a particular disaster scenario. The resiliency curve or the frisality curve relates to the adversarial variables such as whether intensity to failure probability of the individual component. For example, in this figure, the overall line failure probability is shown probability is shown as a function of wind speed. Then, we define a metric for measuring the resiliency score of a system, which could, which we call as risk index, where you can see the PK is the probability of the occurrence for the K scenario and the classification of its impact or the frisality associated with it. Now, the major objective is to optimize the operation of the grid in response to the adversarial scenario. Here, our goal is to deliver the maximum energy to the load, maximum energy or power to the load given the limited generation. The optimal power flow solution are computed at every 10 minutes at fixed interval, which is 10 minutes we have taken to accommodate the uncertainty such as the time varying criticality of the load, the loss in the degeneration, the topological changes and the adversarial event. Now, so here you can see we have the algorithm here for resiliency oriented operation where we assess the threat vector and identify the time varying failure probability of critical infrastructure and predict the time varying priority of critical load based on the scenario and we run the optimization of power flow and then identify the load which the load which needs to be cut out. Here, for a simulation we have considered a co-simulation environment comprising of comprising of both distribution and transmission system. We have taken IEEE 24 bus reliability test system at bus 19 and 3 and we have connected IEEE 13 bus system. We have taken IEEE 13 bus system to do the study. We use the MATE power tool in MATLA for the transmission system modeling and we use the open DSS for the distribution system modeling. Now, we define a dummy disaster scenario, like we define a natural disaster scenario where both the transmission and distribution needs are affected. We assume fault at fault and tripping at some buses loss of certain distribution generation and load. We assume that the load at bus 20 as you have seen before is the time varying critical load where the critical nature of the load for scenario is defined by the following weight function. Here you can see the T, T is the time at which the event or the disasters get initiated and alpha is called, we have taken alpha as 1000 is the time constant of the exponential decay function. Now, for modeling the frisality of the grid due to the disasters we consider the frisality curve for a tornado as shown in the earlier. We assume the distribution feeder at bus 19 has overhead lines that get affected or faulted or fault happens over there with failure probability defined by the frisality curve. Now, this graph shows the value of the objective function corresponding to the optimal power flow at every 10 minutes. This is just a simple test scenario, we are working on creating on some test cases, which resembles to the real world scenario, so that we can enhance the resiliency study. So here you can see the blue line shows the values of the objective function, we keep the criticality of the load fixed, where the lines, the grid. So, here we vary the criticality of the load dynamically. Thank you. Outstanding sensation Shiva. As we wait for questions to come in, I'll have one that I'd like to ask for myself, where do you think this, where is the direction of your research going in the future. My future direction of the work would be like once I'm done with this time very criticality and then we will consider the statistical distribution, and then we'll see like will take more practical cases like this is just like test cases which I'm doing this experimental setup so next would be like a disaster scenario, and seeing how the resiliency, how we can improve the resiliency over there. And what kind of disaster scenarios are you thinking of the size of tornado, were you thinking of like, like, in case of a hurricane or in case of a tornado. Okay, cool. I was just wondering. I know it's particularly relevant for all of us to live in New Mexico. Think about wildfires and how they influence our energy distribution grid and so on. We'll probably look at that too. All right. We have a few minutes for questions and if you can always ask questions. At the end, if we have time as well. We'll give it about another 30 seconds and then we're going to move on to Jesse. Once again, thank you so much for your presentation. Thank you. Thank you, Shuba. We really appreciate your, your talk and your contributions to the New Mexico smart grid center so thanks so much for your time and energies today. And next I'm going to introduce our next speaker and I will have a chance to try and say his name my second chance to say his name correctly. So I'm really pleased to welcome to the screen, Jesse Kachmersky, I'm almost getting there, who works with Dr. Cheney Sherman here at UNM, and Jesse I'll let you take over the screen in the video. Thank you very much. Okay, so what's nice about the, the previous presentation is that it actually kind of segues really well into the type of work that I do here at the economics department specifically for the smart grid center. So, let me just move the panel of people, because it's covering up my slides. There we go. Okay, so my name is Jesse Kachmersky. I'm, you know, working out of the Department of Economics here at UNM, a third year PhD student. And I am here to present the second survey installment that we've done for this, for this EBSCOR project. So, the title of this one is the consumer acceptance and demand of microgrid installations. Specifically, we're focusing on the four corners so we've got Arizona, Colorado, Utah, and New Mexico. And this is a dimension of this larger project that is often ignored, right, so it's very common to build the systems required for distributed feeder microgrid. But then there's the other aspect of it's not always just, if we build it, they will come, they needs to be desired, right. And so, in, that's what I'm going to show you guys what we're looking at around here. Okay, so microgrids they become a staple right in this push towards grid modernization in the United States. So, when you talk about grid modernization in the United States, there's a lot of things that always come up so the first thing would be the smart grid. And then that usually entails some sort of microgrid installation, and then also there is the demand response at program aspect of those. But the question kind of remains is who's going to pay for these microgrids especially if we're going to be talking about community resilience so lots. I would say quite a few of these microgrid installations. They are either done commercially so these are a large facilities that will install them so that they can keep up time. If there are any issues with the grid or say they draw too much electricity. That sort of thing. But then, as we look towards grid modernization in the future, especially areas that have low reliability or are at increased risk for resilience issues. The microgrid becomes a very common solution that's often proposed, but the costs associated are very difficult to measure because each microgrid installation is built specific to the needs in mind of that community, right. But what it at the end of the day, what we see a lot of is the most obvious mechanism for this is that the electricity electricity consumers themselves would be the ones paying for the microgrid. And this happens a lot anytime there's large infrastructure projects, those costs often get passed down through increases in the customer charges on the electricity bill or just slight increases on electricity generation costs in general. And that's usually approved through the public public regulatory commission of said state or area or region. It really depends a lot on how how that electricity provider is managed, you know, are they an investor own. Are they a municipal that sort of thing. It really depends a lot on that. But so what we're going to do here in this the survey is we're going to analyze specifically and distributed feeder microgrid installation for community resilience and liability in the four corners. And so we figured that it's really important that customers have a say in this decision to introduce a microgrid into their grid. So they're my in what they might think it's a great idea microgrid so might be a great idea, but I don't want it in my backyard. There's that that sort of aspect to it too. And so really what I'm trying to do in this survey is illicit though what drives the decision making process to to fork over some cash to for these these microgrid installations. So consumer perspectives actually on microgrid installations have not been researched in the US specifically what drive those decision making processes, and then also what is the median amount of money that a consumer would be willing to pay it just hasn't been done. And so this is the first study of its kind to do that. And because it is the first study of its kind to do that it gives policymakers and industry professionals a metric on what people in general are willing to to pay for these so they can try to tailor microgrid costs for that. And so the premise of a microgrid. I think this is a very technical audience so I'm not going to go into too much depth, but the distributed feeder microgrid system. It's able to island itself from the utility grid and provide electricity and distribute electricity to its, its community. When necessary, basically, so I'm just going to leave it at that it's a lot more complex but that's basic that's what we're trying to evaluate here how much do people really want these. And so the decision making process it's not as simple as saying okay yes there's a resilience increase or liability increase. And so I'm going to just say yes, it's not exactly that simple, because think about people who live in Albuquerque, who have incredible uptime for electricity. We're talking about, you know, never almost never having electricity outages and if you do it's going to be, you know, maybe like 30 minutes at the maximum end. So the, the safety and safety scores and stuff like that here in in urban areas in the in the Southwest specifically are very high, because we're not as prone to natural disasters as other areas of the country. And so the desire or the need is not necessarily there. But there are ways to educate and inform individuals so that they're, we can try to elicit their actual willingness to pay. So we asked these consumers. Would they be willing to pay for the microgrid and that decision making fun process it's actually a function of a lot of things so anything from financial burden so how much do they have to pay. That's going to be a very significant impact on whether they they agree to pay for the microgrid ideological concern so political ideology we see often has impacts on decision making processes and then attitudes and preferences and those that's a very broad term that I left there, but that can be anything from their attitudes and preferences towards pollution from electricity generation to things like they don't agree that the consumer is the one who's supposed to pay for it, you know there there's lots of their attitudes preferences that fall under the ideological banner. And then there's the socio demographic banner so things like how educated individual is how much income they have actually has a large effect where they are so rural urban divide is actually what we find to be a very significant factor in trying to do this analysis is that people at the end of the line are willing to they consider microgrid installations as a as a strong as having strong potential and solving their their reliability outed issues, things like that and versus people as I had explained earlier that that are in the urban setting that don't really have these sorts of issues and other socio demographics. And then there's the desire for community need do does the community actually think they need one. And then there's also an individual need so I might want the microgrid because I know that a certain part of my community could benefit from it, but I don't specifically need one, or vice versa. And then the information so our people fully educated on the microgrid what its benefits are. What's it what its costs are, excuse me, what it can be used for because when you say microgrid, it's such a such a broad word, right. Microgrids very significantly and from that that function of those processes, you can you can elicit a yes no or not sure response. And so we did actually conduct a survey we're still conducting a survey that's why this is the results here I should have mentioned earlier preliminary. So they're not this isn't a complete study, but we started serving people in September, and we're currently surveying individuals, we have about 3500 responses at the moment from the four corner states. And we're trying right now to get up to 5500 responses. That's what we pay for. We're trying to work on that at the moment. The respondents there, they're meant to be regionally representative so we were sampling people based on education role urban divide population statistics between the four states. Things like that income education, I think I already said education. Gender, all these things matter. So we're trying right now are most difficult where we've already got our easy responses out of the way you know the urban high income, that sort of thing we're now we're having issues with low income rural individuals getting those responses. And so these types of responses are collected through Qualtrics. It's a Salt Lake City based company. And so they're working on that collection right now we were close hand in hand trying to do that. And then what we do in the surveys we fully inform them of the benefits of a distributed feeder microgrid. And I'll get into what exactly we what type of information we provide them in the next slide, but we actually also ran an experiment in this survey so we wanted to determine how do people's desire or willingness to pay for such a microgrid actually differ, depending on what type of infrastructure they receive. So if they receive direct benefits or indirect benefits from the microgrid. So, direct benefits that experiment was basically that during the information process we ask them, would you actually go for this program, knowing that the microgrid would provide electricity to your community, and then support the critical infrastructure in your community during high stress events. And then indirect benefits is where we basically skipped that and said that it was installed in a nearby community, but in times of grid stress that microgrid being on the larger grid could reduce the probability of outages in your in your community. You basically have two very different benefits structures that we're presenting to the respondent. And what we want to do is measure the difference between the two as well. And so the information we provide them. We run them through a battery of information. We tell them what a microgrid is. Why do we need a microgrid, especially in the Southwest. We give them economic resilience reliability and environmental benefits, and I noted a caveat on that environmental benefits, because for some reason, the general assumption right now is that microgrids are renewable, or they have some sort of environmental benefits to them. But in reality, only 20%. I think there's a latest statistic I could find only 20% of microgrids in the United States are renewable only. And that's the type of information we provided them we said, we didn't say that it was going to be renewable or not what we did was give them that percentage. And we allowed them to mentally create their own adjustments from that for costs, we informed them of the cost of producing electricity at a microgrid on average. I can't remember where that source was from. But it is a it is a national lab that did a study on that. And we used that and compared it to the average price of electricity per kilowatt hour in the Southwest. And that's where we left it so. And then for because you can't specifically say okay this is the average cost of a community microgrid because they're so they vary so much. And then, for examples, we gave them examples so we told them about how microgrids provide critical infrastructure supply and at the Denver National International Airport, and then also how a specific facility in in rural California was able to insulate themselves from rolling blackouts over the last couple of years, because of their, their installation from their microgrid installation. And so, then we asked them a basic a basic referendum style question, you know, taking into consideration your desire for this microgrid, as well as your current disposable income. Would you vote for it if it costs you if it increased your, your customer charge by a certain amount of money for 24 billing cycles, right so this is a two year commitment that they're asked to to vote on. And then they're given a randomly distributed amount of money that's actually based on the four corners average electricity bill. And so, there are eight levels in which a response it could be randomly chosen, these are chosen on a uniform distribution. And each person would see one of these right so that could be on the rounding column, 10 cents 50 cents added to their bill for 24 cycles but some people would even see $17 added to their bill over the 24 cycles. So this is, and at the top you see the average bill in the four corners for the entire year is $86 and 97 cents. And we built it off of this structure because we want to basically with the type of methodology we use. It's, we use a logistic regression to estimate our point estimates, and then we use median willingness to pay calculations. Traditionally, that's a, that's from Cameron and James, but we use a hub and McConnell method from 2003. That's just the citation for you. But this exponential increase is interesting because we're going to be able to see how people respond in their votes, depending on how much money that we asked them to pay for. And you see that here so what you would want to see is exactly what we see in our responses so people who voted yes in the top left panel. They voted yes overwhelmingly when they saw really low bid offers or offered payments right. So if the cost was within like 10 cents five cents whatever people were saying yes you know that's not a big deal. That's not a big deal the more you asked people to pay the last people said yes, the opposite should then be true by yes so the opposite should then be true in the no vote so people who voted no overwhelmingly probably did so because they were seeing very high responses you see that with the upward slope there. Also these lines aren't fitted in any way I just threw these on there and PowerPoint to show a directionality so it's not that perfect. I'm not sure about if people are actually not sure. There should be no visible distinction in a trend basically in offered payments by vote type and what we do we do see that so that's that's a good little check to see that people are actually saying not sure because they're not sure, which is which is a nice little check. We asked them initially before that question that you saw we asked them initially if it was no cost to you, would you vote for this program. Would you vote to have an installation installed about 57% of people said yes, almost 30% said no, or I'm sorry not sure, and the rest said no, no, but when we asked them to pay the people who weren't sure became a lot more sure. And now we have about 30% of the people saying no, and less people saying yes, and that's what you would expect from the surveys that styled in this manner where you asked them to pay now, then people kind of drop off. And so, from running a logistic regression I don't use any point estimates in here because we would just take all day to explain them all, but some key takeaways from what we find is that the more so the more that the offered payment is. So the more we ask them to pay the less likely they're going to say yes. So all of these are in relation to the likelihood that they're going to say yes. So we actually recoded not sure as knows we didn't give them the opportunity to be either yes or no, we just recoded them as not sure, or I'm sorry no so these are extremely conservative estimates but And so, if they fell into the category of respondents who were going to receive direct benefits from the microgrid, there were much more likely to participate in this study. I'm sorry in the to vote yes on the referendum, and then the utility. Do they think that the utility actually has their best interests in mind this is a really good question because you want to try to elicit do they do the people who are responding actually like trust that really their provider right do they think that the providers doing the right things for them that people who agree with that actually didn't see participate in the program a lot more. And then, whether they had a concern for pollution from electricity production was also positive and statistically significant. This is interesting because it sort of creates this, this slight mechanism between, do they think micro grids are going to lead to a reduction in electricity production. This isn't necessarily true, and people who bring a lot of people in urban environments actually receive electricity from larger, you know, plants that are very far away right like we get a lot of electricity from power plants and things like this, and they come to us, but what we're talking about is adding a microgrid to your, your community, or it's going to generate electricity in your area. It almost maybe hints at people not fully understanding, we're taking into account that voting yes for this might mean that you're going to have a natural gas or diesel generator two blocks from your house so you know, so people that wasn't very clear. And then, so this cost sharing ideology we asked them do you think that electricity can customers should bear a burden in sharing cost for for this microgrid for like infrastructure upgrades, people who did agree with that and we're much more likely to participate. Political ideology, this is negative because we measured it on a scale where from the mean, are you more conservative so more conservative respondents who self reported being more conservative, but we're less likely to participate in this, this microgrid installation. People who work from home post cobit, much more likely, and that's likely because of the proposed reliability benefits of community microgrid, and then income is just a check that we do. If you have more money, economic theory, a very shallow economic theory should suggest that you you're able to have more disposable income and more likely to participate. And that was true. And so, from those estimates we can actually calculate what the median willingness to pay is for an electricity customer in the United States, or I'm sorry, in the four corners. And so, if we don't include any of the covariates that any of these predictors that we have on this slide. It's about $8 and 21 cents per person per month for 24 months for direct benefits those are for people who are going to receive the direct benefits for indirect you would expect and it is what we see that there is less willingness to pay. So by $2, which is a which is a large amount. And these are both highly statistically significant. At the net, I think at the 99% confidence interval. So very strong results. And then on that, that right most column is going to be the total over the 24 months. And then if when you add the actual covariates here that we're used to predict. And this is more getting into the weeds of how we do willingness to pay calculations. It doesn't actually change that much from the original estimates. And this is calculated again as I said earlier using Habin McConnell's linear willingness to pay methodology. And so, these are preliminary results these are just a sub sample of the total results that we're, we're, we're trying to get. But our future research, most like that is very simple, very shallow. And that's also what we have actually going on what we're going to be doing a lot of this heterogeneity analysis so looking at regional. So how does Utah willingness to how does willingness to pay differ from being in Utah or Arizona or New Mexico. What about rural urban divide so that those those rural individuals who suffer more so from outages and things like that. We're going to be able to pay more or less. I'm not sure because being rural is oftentimes more correlated with having less income so we're going to be digging into that some more. We're going to also be looking at the differences of willingness to pay amongst various demographics, and then also political ideologies. And then we also have a lot more to look into in the survey I mean we have things like how frequent was their mode, how frequent our outages for them what was their longest outage have they ever been affected by wildfire. And then we have an entire wildfire evaluation section that my colleagues going to be working on. How does the response change when they know that the willing, the microgrid is for wildfire mitigation, things like this. And then, yeah, alright so waiting for the remaining survey responses. And then we'll be doing ourselves. I'll be doing a reassessment of willingness to pay because I asked a question in the survey that said, would your answer have changed if the microgrid is guaranteed to be all renewable. So I'll be able to take that response and recode responses and see, you know, what does the willingness to pay look like, if it is all renewable. Right. It'd be very interesting to go into that. And then, as being on the New Mexico smart grid center we plan to do a microanalysis of New Mexico so we oversampled for New Mexico. And so we're going to be doing a microanalysis of that. With that, that's the end of my, my presentation and I'll be more than happy to take questions so awesome. And then your presentation. So, I will run in a little short on time, but if we get any questions in the Q amp a box, I will be happy to ask them. A question that I got from another source is how has coven impacted your research and are you seeing this reflected in the survey. I'm not seeing it reflected in the survey responses but you feel like it will have a noticeable impact on your potential data. That's a really good question. And we have lots of controls for that because we didn't start running this survey until September so we had a lot of time to really build a survey around coven. So we have questions asking people respondents, you know, were you financially impacted by coven 19. Are you working from home more now that coven 19 as you know as a result of coven 19. I already said financially impacted, but that's the big one is the financial impact of coven 19, because that we're asking people to accept a higher financial burden. So that that question really boils down to income. And so we have questions in the survey, and we're able to estimate how those impact responses so we can run those in the logistic regression and see if they're financially impacted by coven. Does that change the likelihood that they're going to vote yes, or no in the program. I have yet to do that analysis at the moment. But that's a really great question. It's there. And we're going to go into that a lot so fear fear no more. We'll take care of that. So I think we're going to bring something on. Thanks Brittany and thanks Jesse so much for your talk it's really exciting to see the kinds of results that you're getting it's, it's, I, as a social scientist myself I'm glad to see that that people are included into these overall models and seem super important. And just in your talk that you, your senses that people don't really understand what a microgrid is. And I just want to mention that part of the smart grid center actually we're doing some work with explorer in helping to promote that idea, and we'll be developing activities that get that get deployed at museums around the state to teach people about micro grids, and then ultimately explorers going to mountain exhibit at their, their new teen center about micro grid. I think there's some opportunities actually to take some of your findings and, and, and put them into those efforts. So, well we really appreciate your, your time and energy on on the smart grid center research and thanks so much for being with us today. There is a question that just popped up in the chat. Please go for it. Give me just one second I'm going to take a read. You can go ahead and read that out loud. Yeah, so I discussed this political stance quite a bit in what we use to kind of control for responses and the question is do you think that the transition to a more renewable friendly presidential administration will change responses as the national discourse changes. That's a, that's a really good question because I just want to permanently unmute myself so I don't have to hold the spacebar, but Wow, that's a, that's a heck of a question actually. It's not as simple as one would think because the assumption that micro grids are entirely renewable. It's just not true. And, but if it were true you know if, if, if we do start just, you know, using only renewable electricity that could change responses entirely and that's kind of a larger focus of my, my, my future research coming up here we have questions for that like I explained earlier. But political ideology and its impact you know that's usually a response that's associated with rhetoric and then also the the mechanism between political ideology and micro grid installations, it might not necessarily be because of renewable energy or anything like that it might actually be much more so that political ideology is correlated with being fiscally conservative. And so they might not want to invest in community projects as a result. So there's a lot to unpack there. And I actually think that the renewable friendly discourse is going to be a very small portion of what drives political ideologies relationship with micro grid installations. Thank you so much. Yeah, it seems like this is an especially interesting time to be doing this work as things are, are changing in the environment. So I would like to introduce our next speaker for today, and that is Ali Gorashi, and he works with Dr Ali Bidrum at the University of New Mexico and I'll let Ali unmute and share your screen. Hi everybody. I'm Ali. I'm a PhD student at UNM. I'm PhD student at UNM. And today I'm going to present some part of my research that is supported by enemy PSCOR program. I'm going to talk about the cooperative dynamic power balancing and smoothing in a photovoltaic hybrid energy system using multiple reactive agents. Firstly, I provide an introduction to the topic after that. I discuss about the power management systems in PV hybrid energy source system after that I provide a short description about the cases study system and after that I propose a distributed hybrid control strategy in this work. And then I evaluate the performance of the system using computer simulation and finally I will conclude this presentation. So the first question is that why dynamic power balancing is important? You know, it's a basic rule in power system that the amount of power that is generated must be always equal to the amount of power that is consumed. Otherwise, we will have a blackout. This is more challenging in PV power systems. Please assume diesel generator in this case you can increase the fuel on the engine and you can increase the power generation and you can reduce the full to reduce the power generation. So you can easily control the output power and balance the generation and the load. But in PV power systems, we have not full control on the power generation because you know the power generation is related to the weather condition and time of the day. So in the systems that we have no access to the utility grid, for example, as human isolated microgrid system, we need additional devices in order to balance the generation load and these devices are energy storage systems. Here you can see a very simple explanation of how energy storage systems works in a PV ESS system. In a very simple term, we send the difference between the generation power and the load power which is called the net power to the energy storage system. As you can see in this figure, when the PV power generation is higher than the load power, the output power of the energy storage system is negative. So the energy storage system plays the role of the load for the system. And when the load is higher than the power generation, when the load power is higher than the generation power, the output power of the battery is positive. So battery plays the role of a generation unit. So with this simple strategy, we can balance the generation load in a PV ESS system. So in dynamic power balancing, the sampling time may be less than one second or around in this scale. As you know, we have lots of fluctuation in the load and also we may have lots of high frequency variation on the output power of the PV. So in this case, the net power that is sent to the battery energy storage system has a high frequency variation and this high frequency variation sometimes is problematic for us. In addition to that, the battery energy storage system that are widely used for power balancing, for example, lithium-ion batteries, they have a slow dynamic response and sometimes they cannot track this high frequency variation. And sometimes in isolated system, we may have some slight deviation in the voltage and frequency. In addition, these batteries have limited life cycle. There's a number of times that we can charge and discharge these batteries limited. So when we send this high frequency variation to the battery, the battery needs to frequently charge and discharge, which reduce the lifetime of the battery. In order to avoid this problem, in HES technology we use a supercapacitor that works in tandem with the BES in order to improve the response of the system. Here you can see how hybrid energy storage system works. As you can see, there is a power management system which is responsible for power allocation between different energy storage units. As you can see, it sends the low frequency variation of the net power to the battery energy storage system so it will increase the lifetime of the battery. In addition, the high frequency variation will be sent to the supercapacitor. The supercapacitor has a high power density and has a fast dynamic response so it can easily track this high frequency variation so the voltage and quality of the line will increase, especially in isolated system. There are different ways in order to design power management system in PV, hybrid and energy storage systems. The rule based methods are widely used because they have a lower computational complexity and they're really suitable for real time applications. But in these methods, usually there is a centralized supervisory controller which is responsible for decision making for all the primary controllers inside the system. But using the centralized architecture has two major drawbacks. The first drawback is that a single failure on the supervisory control, on the centralized supervisory control will lead to a system collapse. In addition, if we want to design a very flexible and adaptable system, we need to design lots of operational mode for the primary controllers. And in centralized architecture, it can be very tedious test for the system designer to do this because it requires a very complex logic for the system. In addition, implementation of this complex logic using a microcontroller may not be suitable and may not be applicable in some real time application. In order to avoid this problem in this work, we provide a distributed management technique in using a multi-agent based control strategy. Before I introduce our proposed approach, please look at the cases study system. The cases study system contains four modules. One of them is a PV power generation module, another one is a battery energy storage system, and also there is a supercapacitor module. We have also a load module. The load module is an isolated AC microgrid that the portion of its power demand is supplied by the PV hybrid energy storage system. The structure of our multi-agent based control strategy is that in our approach, each module is considered as an intelligent reactive agent that can cooperate with other agents and change its operational mode and dynamic behavior with respect to the condition of the system. The cooperation of these agents provide a global pattern of organization that forms a dynamic power balance in the small safety system. Here you can see the hierarchical control structure of each agent. As you can see, each agent has also a data acquisition and information processing module. Each module received data from communication using, through communication with other agents and also from local measurements. They process the information and send them to the hierarchical control system. At the top level of this hierarchical control system, there is a supervisory controller which is responsible for dynamic decision making and selecting the operational mode of the low level controllers. There is a power charge controller which is responsible to calculate the reference current in order to obtain the objective of the system at each operational mode. Then there is a primary controller that calculates the duty cycle of the power electronic converters in order to track the reference current. Here you can see the hierarchical structure of the PV module. You can see the PV module can work in maximum power point tracking or it can reduce its power generation and works in low following mode based on the condition of the system. One of the novelties of our work is that we let the PV module to cooperate with other subsystems, especially the supercapacitor and the cooperation between PV and the supercapacitor. We will see that it will increase the reliability of the system. Here is the hierarchical structure of the supercapacitor. As you can see the supercapacitor can work in fully charged or MPT mode. In this case it is in idle mode and also it can work in normal operation and in this case the high frequency variation of the net power will be sent to the supercapacitor. Here is also the hierarchical structure of the battery and energy storage systems. As you can see the battery and energy storage system has a smooth charging module. Typically, as you can see in the top figure, the typical smooth charging method is illustrated. In this method there is a low pass filter, so filter the high frequency variation and send the low frequency variation of the net power to the battery and send the high frequency variation to the supercapacitor. But in our proposed approach we use a controlled feedback in order to provide a cooperation between the supercapacitor and the battery. As a result of this cooperation the battery can control the state of the charge of the supercapacitor so it can prevent the supercapacitor to frequently become charged or discharged because the supercapacitor has a very low energy capacity. Using this technique you can see that the minimum required size of the supercapacitor will reduce. In order to simulate the behavior of the system we need a dynamic model of the system. You know each agent has a hybrid dynamic behavior. The top level controller, the supervisory controller is a discrete even dynamic system and also we have time driving dynamics which are the dynamic of the controller, primary controller and charge controller and also the dynamic of the physical system. So each agent is a hybrid dynamic system. In addition we have lots of interdependency between the discrete and continuous dynamics of the agents. So in order to model all this concurrent dynamic behavior of the engines in one framework we use an input output hybrid automaton which is a very suitable framework for modeling concurrent complex hybrid dynamical systems. So here is a case study system that we simulate this behavior. As you can see there is a PV module that has 100 kilowatt power capacity. Also the battery energy storage system has 100 kilowatt rated power. And the supercapacitor also have a seven seconds charge time. Here we define a term which is called the total power or PT as you can see, which is equal to the sum of all auto power of the modules. And it must be always equal to zero in order to make the balance between generation and load. We use this term in order to evaluate the power balancing performance of the system. Here you can see the performance of the system during a normal operation. As you can see the net power has a high frequency variation. These are the blue lines. But the power that is sent to the battery has a smooth variation. So the power of smoothing performance is very good. As you can see, you can see the green line which is the total power and it is always equal to zero during the simulation simulation interval. So the power balancing performance of the system is also great. Now here we examine and evaluate the performance of the system during a sudden variation on the load. As you can see when there is no cooperation between the PV and the supercapacitor, after a sudden loss of a major load, the supercapacitor immediately becomes full charge. And it cannot participate anymore in power balancing and smoothing performance. So we will have an unbalanced power generation load for around 30 seconds. But using our technique that has a, and in this technique we have a cooperation between the PV and the supercapacitor, you can see that when the PV feels that the supercapacitor is going to be full charge, it's reduced its output power and exactly for the amount of power that the supercapacitor is supposed to absorb. So the system can maintain its balance during a sudden loss of a major load. So you can see that using this technique we increase the reliability of the system during a sudden change of them during a sudden load change. Here also you can see the performance of the cooperation at the advantage of the cooperation between battery and the supercapacitor. In this experiment we use a supercapacitor that has a 100 watt hour energy capacity. You can see that with this capacity, the supercapacitor, if there is no cooperation, the supercapacitor becomes full charge because there are some low frequency power sent to the supercapacitor as a result of non-ideal filters. But when we provide the cooperation between the battery and the supercapacitor, the battery prevents the supercapacitor to immediately charge and discharge. So the system can operate in a normal operation and the PV don't need to reduce its power generation to maintain the balance. So we have more power generation. In addition, as you can see in the second experiment, we test the performance, we test the advantage of this cooperation during a sudden loss of a major load. As you can see, when we have no cooperation, it takes around six minutes for the supercapacitor to come back to the normal operation. So it takes six minutes for the PV to come back to the maximum power point tracking. But using this cooperation, the battery immediately discharge the supercapacitor because it has a control on the capacity of the supercapacitor. So the PV can come back to the maximum power point tracking after 30 seconds. So as you can see, the efficiency of the system increased by reducing the required size of the supercapacitor. Now to summarize the results of this research, I can tell you that we designed a complex logic for the system. The logic of the agent is not very complex, but the logic of the overall system is very complex and it is not easy to design this logic using a centralized controller or it can be a very tedious task. In addition, so as a result of this, we provide lots of cooperation between the modules inside the system. As you see the cooperation between the supercapacitor and the PV increase the reliability of the system and the cooperation between battery and the supercapacitor increases the efficiency of the system. So we reduce the complexity of the system design, we increase the reliability as well as increasing the efficiency of the system. Thank you very much for your attention. Please ask me if you have any question. Thank you. Yeah, that's great. Great presentation. We do have a few questions actually. So we've got two for you. Did you consider the line? I need to stop sharing. Okay. You ready? Yes. Did you consider the line impedance of solar PV, best and supercapacitor? No, no, I didn't consider this impedance, but in our recent research we are trying to make the model more complicated. For the supercapacitor, we consider the supercapacitor is ideal and it doesn't have any power losses inside it. In order to simplify the mathematical model of the system because the model that we use is very, very complex. So to reduce the complexity we didn't consider the loss of power inside the supercapacitor. Awesome. We've got another one. Would there be any performance impact on the system if the data communication rate is different? Excuse me. I couldn't understand. Could you please repeat your question? Would there be any performance impact on the system if the data communication rate is different? Yes, but you mean if the data communication was, if you have a problem in the communication, you mean? I think so. I'm actually going to really quick unmute the person who asked the question and they can ask it themselves. Would you be any effect on the control performance of the supercapacitor source-based converter if the double frequency components? Would there be any effect on the control performance of the supercapacitor based converter if the double frequency components? I'm not checked it, yes, but I think it doesn't have any effect on the control performance of the system because the mathematical equation that we use and the dynamic equation of the supercapacitor doesn't have any direct impact on the operation of the supercapacitor. You know, there is a question that data communication rate is different between agents. You know, they have different, the agents, the communication between agents, the signals that the agents sense to each other is related to the power of their, if each other. So the communication rate of the agents is not very high, but all the agents, we use the same communication for the top level controllers. The low level controllers of the agents just communicate with their associated higher level controllers and not cooperate with the, for example, agents and agents low level controllers. So the high level controllers, they receive the data of the other agents. So, in this point of view, they have the same rate of communication rate, but inside the system we have different rate of communication because it's related to the level of the controllers. Fantastic Ali. It looks like you've got a lot of really great things going on and we really appreciate you presenting today, but also contributing to the research work of the New Mexico Smart Grid Center. Thank you. And next I'd like to welcome our final speaker, Anju James, who is a student at a New Mexico State University, and working under the direction of Dr. Jay Misra. And Anju, I'll let you take the screen and your. Thank you. Hi, I am Anju from the Computer Science Department of NMSU. And I'm working in communication side of the Smart Grid. Today I'm going to talk about the quality of service software, NDN based network architecture for Smart Grid. And this discussion includes what is NDN, why NDN is required in Smart Grid, our proposed architecture and some simulation results. NDN is an information-centric networking and it supports reliable communication architecture for Smart Grid in terms of scalability, protocol interoperability, security and quality of service. And we enable QoS in communication network by classifying the traffic based on priority, implementing multiple transmission queues and using token bucket for traffic control. So initially we will look at what is NDN. NDN is a communication architecture and NDN stands for Named Data Networking. Currently we are using IP as our communication architecture in which we use source and destination address for the communication. Named Data Networking is an alternative for currently used IP architecture and proposed future internet architecture. The basic fundamental block of this Named Data Networking architecture is the node and we can configure each node into consumer, router and producer. As you can see in this figure, this is the very simple communication topology of NDN architecture. Consumer or subscriber is the person who is requesting for the data and producer is the person who is providing the data. The data request is known as interest and the producer will respond back with the data requested. The intermediate router is responsible for the data or interest packet forwarding. As the name resembles, NDN manages the communication between entities using the name of the packet. In IP, we know it is based on IP address. Here we use the name of the packet. And there will be a NAT packet which is negative acknowledgement sent to the consumer in case the router could not forward the interest to the producer. And we can use the namespace according to application requirement and it will be hierarchical instruction. We will discuss about it in coming slides. For the simulation purpose, NDN offers its own simulator which is known as NDN SIM and it is based on NS3 network simulator. First we will see what is in order. In order of NDN system includes several modules. First is content store which is a cache which supports the data reusability at edge routing. Next is pending interest table. As the name resembles, it is the table which stores the information of the interest. Next is forwarding strategy which is similar to the routing protocol in currently used IP architecture. It offers, by default it offers multicast best route and some other strategies and we can create our own custom strategy. And there is a FIB which is forwarding information base which is equivalent to the routing table in IP protocol and it stores the destination address and path to it. Next we will see the NDN architecture. So first we will see how an interest packet is forwarded from a consumer to the producer. The interest will be generated from the consumer and the interest packet will be forwarded to content store. This is a single nod in the communication architecture. Once the interest packet is received in this content store, it will check whether there is an entry as same as the interest. That means that particular data is requested by someone else before and it's already cached in the content store. If it matches, the data will be returned and communication overhead is reduced there. If there is no matching entry for this particular interest, that interest packet will be forwarded to pending interest table. We will have an interest table with the interest name, incoming interface and outgoing interface. Sending time. So we will check whether there is a matching entry in the pit table. If there is a matching entry, that means someone else already requested for the same interest and waiting for the data. So we don't have to request it again. We simply add the incoming interface information in this table and we will wait for data to come back. If there is no matching entry, then based on the fifth table information which is the routing information and the forwarding strategy which is the routing protocol. Based on these two elements, we will be forwarding it to the producer. Next is how data is forwarded from producer to the consumer. Once we receive the data in a node, we will be checking the matching entry in the pit table because here we don't have any IP address. So we have to send it back to the same interface which requested for that particular data. Those information will be stored in this table and if there is a matching entry, the data will be multicasted to all the interfaces which requested for this particular data. And after that the data will be cached in the content store and pit entry will be removed. Sometimes when this data is available, there will not be any matching entry in this pit table because it is the timeout happened. RTT is the round trip time and if there is a timeout, in that case also this particular entry for interest will be removed from this table. This is how the NDN architecture works. We will see an example. So here is one user and he need a video from CNN. So he will generate an interest with name CNN slash video and it is connected to router D and the fifth table which is forwarding information base. In this table which will have entry for that particular producer and the path to it. This will be in ascending order of priority. So first we will choose for A. So if the interest is forwarded to A interface, that means it can reach to the data. If A interface is not working, it will look for B and if A and B are not working, it will go through this path. So this is the hierarchical structure of the name. Next we will see what is the importance of NDN in smart grid. Smart grid means power system with communication. So here we have presented one diagram which uses IEEE 39 bus system and communication network is embossed on top of it. So you can see multiple entities here PMUs, PDCs and wide area controllers. These entities how to communicate each other and for that purpose we are using this communication topology and this blue links and this blue box represents the communication network. This blue are the routers and here you can see an interest name. This is an example name we can use whatever name we want based on the application requirements and here we use IEEE 39 and here is one for priority class. Based on this information it can be high, low or medium. Based on that information we are classifying the packet into three different priorities. PDC, measurement, PMA. This is just an example we can have whatever we want. So the currently existing IP uses IP address and it has its own drawbacks. It does not do well in device heterogeneity, protocols and standards, interoperability, application QoS requirements and security. In this particular research we are focusing on the quality of service requirements of the application. In our proposed architecture we are introducing a new strategy which is QoS strategy and it is derived from the existing default multicast strategy. In this model we are classifying the packet into three different priorities, low, high and medium. The protection messages which requires low latency and high reliability is termed as type one traffic. Control messages which requires high reliability is termed as type two traffic and others are termed as best effort traffic or type three traffic. And we will have multiple Qs and each Q is based process based on weighted fair queuing and there is a token bucket. In each node of the network we will have this system. So this is the more detailed view of the multiple queue implementation and token bucket. We will receive a packet with the priority information in it and based on that priority value from its name we will classify it into high, medium and low. And these queues will be processed using weighted fair queue algorithm and to control the traffic flow we are using token bucket. Each queue will have its own token bucket and token generation rate will be different for each queue. Even though this next is the packet from high priority queue and there is no token available for that particular queue means this packet will not be dequeued. So that is how we are controlling the communication data flow. This is some of the simulation results we have included we did this experiment in IEEE 39 best system and we are comparing baseline Indian ICASM and ICAP. So ICASM is our previous work which focus on the multiple available paths and ICAP is our current proposed system which uses multiple queues and token bucket. Baseline Indian is the original Indian framework which does not support quality of service. In our simulation we used we configured the network into 27 routers, 10 wide area controllers, 12 PMUs and 2 PDCs. And we can generate this, we can configure the source nodes and produces as per our requirement and we can set packet generation rate differently for different types of packets like as you can see here type 1, type 2 and type 3 hard. Packet generation rate is 90, 150 and 300 packets per second. Also we can configure token generation rate also differently. Token bucket capacity we maintained 1000 tokens for all the traffic flows. If you see the graph our proposed architecture over performed all the other methods like baseline Indian and ICASM. In the case of loss rate type type 1 and type 2 was outperformed but in the case of type 3 baseline Indian is doing better. The reason for this is for type 1 and type 2 traffic baseline Indian uses unicast and we uses multicast that means network congestion will be more. ICASM and ICAP both uses multicast and we will be we are doing better than ICASM and we enable the quality of service also. This is the comparison between IP and NDN. So this is this is this we did as part of the ICASM work which does not support the quality of service. If we see the packet loss comparison, we set up the network to test for 0% network congestion, 20% network congestion and 50% network congestion. We can see that for NDN we don't have any packet loss. Next is the comparison for latency. This is 0% network congestion. This is 20% and this is 50% network congestion. Regarding the latency, we can see that UDP is doing better than next is TCP and last comes the NDN. This is because UDP has the less header size and that is 8 bytes and for TCP it is 20 bytes and for NDN it is 40 bytes. But as the network congestion increases, we can see that packet delivery is getting affected for TCP and UDP but there is no much packet delivery changes for NDN. So if we see this within 30 millisecond, NDN is completing 100% of the packet delivery, UDP and TCP is 75% and 80% perspective. So this is our quality of service of our networking architecture overview. We are classifying the network based on the priority to high, medium and low and we did a lot of other experiments to check the scalability like this is IEEE 39 by system. We also did Monte Carlo simulations in 123 by system, 650 by system and we could identify a relationship between the packet generation rate and token generation rate. Also, in future we are planning to use the machine learning models so that we can dynamically assess the network behaviors and adjust the token generation rate for that. So that is our future plan. Yeah, that's about our research. Thank you. That is fascinating. That's absolutely fascinating because it just shows how far we've come as far as our networking architecture goes. I had a question come in that is a little general but I'm hoping you'll be able to answer it. How long do you think it will take for NDN architecture to replace IP UDP architecture? Is that happening in the next few years? Do you think it will happen? NDN is kind of developing stage so we cannot, we may have to perform the real-time simulations and this is just simulation we need to test it in real-time scenarios to evaluate the system. Maybe in near 10 years it will get implemented. We are just almost at time. Again, thank you for your question and I will turn it back to Celina. Thank you, Brittany. And thank you so much, Anju. We really appreciate your presentation today and it's exciting to see that you're finding results in the simulation platform and also looking forward to seeing things, where things go next with machine learning. Maybe using some of the testbeds that the Smart Grid Center supports at SWTDI at New Mexico State and Mesa does all appear at UNM as well as the microgrid, nano grid at the Santa Fe Community College. So one of the things that is sad about being in the virtual environment is that it's very hard to show appreciation to our presenters but please join me in a virtual round of applause to thank our presenters today. And it's fantastic to see such great research from our young researchers. I'm going to share my screen real quick here. And just do a quick reminder that tomorrow will be the award ceremony related to this New Mexico Research Symposium and so you're welcome to join us at three. All of our events are hosted up out of our website and we will post this recording there as well. You can expect to see it sometime before the end of next week. So if you have anybody that you'd like to share this with, it will be available at the EPSCOR website then. So with that, I think I'd like to call an end to our Smart Grid seminar. Thanks to our speakers. Thanks to all of you for joining us today. And we hope you have a good afternoon. Stay safe everybody. Bye-bye.