 Hello everybody, hi, I'm Scott Hinson. I'm the CTO of con street. Thank you for Joining me this afternoon. I hope that you enjoy this presentation. I am Sorry that for anybody that was looking forward to Doing this in person that we can't but it's the way this goes right now. So We'll do the best we can normally I do a lot of audience interaction. So please Please get those questions in Early and often a lot of times the questions help me make sure that I'm addressing the What people are interested in hearing about? I am in a room full of equipment. So occasionally you may hear things happen in the background I will try not to let that distract me too much But we've got some active tests running for some national labs right now. And so you've got Some equipment that you'll occasionally hear come on and turn off In fact it turned on a few seconds ago So The one we talked about today is a Program that con street my organization did in partnership with awesome energy called shines It is a DOE research program around energy storage renewables getting more solar systems and distributed energy resources onto the grid and a low-carbon way The first a little bit about us con street. We are a 501 C3 non-profit We have sort of a three pronged approach to how we we do Business in the world. The first is is that we do a lot of field research. We have some very Generous volunteer participants that work with us nationwide We install equipment in their homes to monitor electricity usage water usage natural gas usage And we get a lot of data points, okay? We Have what we believe is the world's largest private database on residential energy usage in fact even if we were a Utility we think that we still have the world's largest database We have homes that because of the types of data that we collect Generate ten and a half million data points a day. We collect 2.5 billion data points a day on electricity usage alone And so the the amount of data that we collect is is fairly staggering every second of Every day for these homes for every single circuit. We collect five data points and that is Real power parent power and I'm going to use some electrical engineering terms because that's what I am But we'll get back to those in a second real power apparent power phase angle distortion on the current and the Magnitude of the current and and we do that because when you start putting all sorts of devices like batteries and solar panels onto a grid that already has sort of energy efficiency devices like a modern air conditioning unit or a You know compact fluorescent light bulbs LED light bulbs computer power supplies There are some very interesting interactions that happen with those devices And unless you are measuring at a very high resolution with very special equipment You don't see those interactions and if you just ignore them Bad stuff could happen and and you know a problem is not going to go away because you're ignoring it um and The last Not least is that we have a testing verification lab um And so so between field research where we we put stuff out in folks homes where we've got this measuring equipment And we can see the impact that it makes on their energy usage their energy efficiency How they interact with the the energy water natural gas world around them? Turning that data over to researchers worldwide and and corporations worldwide and then testing individual devices before we put it in the field in our lab We have a full suite of things that we can do for Federal programs private research grants We've had grants from Sloan foundation. We just got another grant from the government foundation So these are private grants that are private entities that are looking to to seed research in certain areas so shines Is a program that we did with austin energy um and the department of energy And uh it had Three major components what we were trying to do with shines Was understand how you could use energy storage and smart connected internet of things devices to increase the grid's ability to host distributed resources distributed generation and lower the cost of electricity, okay and and it's it the cost of electricity is not just the cost of the fuel that it takes to generate that electricity and get it to you There's a whole bunch of other costs. There's there's um Capital equipment costs. There's operational costs. There's all kinds of things And and in a world, you know going back 20 years ago When all electricity was generated at a massive You know relatively small number of centralized power plants You knew exactly which direction the power was flowing on the wires and exactly uh, and you could have um a pretty easy way to to sort of Anticipate what the load was going to be on any given circuit on any given day based on weather That has all changed right now. You've got a system where people have control of thermostats Remotely, but thermostats are learning if they think people are out of the house or in the house They're going to turn things on and off Um, they may have solar on the roof and it may be generating Quite nicely in one area, but a mile away It it is cloud cover and and generation is very low and all of these things mean that Utilities are under increasing pressure to manage these power flows and there's cost To that right and so a smarter management of these things could reduce their costs. They may be able to avoid Upgrades to circuits. There's something called congestion management in the utility industry where they're they are managing How much power is on on those wires at any given point So they don't have to go in and put new bigger wires in right you've you've seen this driving. Well, I noticed this I may be the only one that notices this Our other utility folks Notice this, but I've driven down the road and I'm watching Very expensive Wire towers go in right next to older smaller ones And and either that is because of population growth Or it may just be because they have some sort of constraint on the system that that they're trying to manage And energy storage systems and the proper operation of energy storage systems may be able to avoid that right So that's one of the things that we want to look at. So for the shines program We were looking at three classes of energy storage and sort of three classes of devices Three classes of devices that Are useful First is utility scale storage the next is commercial scale storage so sizes for business utility scale storage is typically on the order of megawatt hours of storage with megawatt peak power capabilities To put that into perspective that will run hundreds if not thousands of homes, right? Commercial scale storage systems are somewhere between 30 kilowatt hours to 150 kilowatt hours sometimes a little larger and maybe 75 kilowatt peak And that's enough to offset some of the production of the business And there's some economic reasons why they might do that and then residential systems might only be 10 kilowatt hours Well And five kilowatt peaks were delivery. Well, they all come at different costs, right? And if you just look at Kilowatt hours of storage versus cost the utility scale systems win every time, okay? But the utility scale systems can't always be put exactly where you need them It's very difficult to sort of deed the land and do all these things So now you can put smaller commercial scale systems and tiny residential systems Exactly at the point where you'd want them And now they become far more useful So there were part of the study was that we were looking at was to figure out if The residential systems had a lower overall cost of energy usage even though their their equipment cost up front might have been higher Okay So what did we install? We installed We we say here that we've got six energy storage systems We actually have a seventh in our lab and then we added a vehicle to grid so out in the field in in in the wild On our participants homes. We had six energy storage systems We also had additional smart inverter systems And we have communications established with every single one of these systems And we have very high speed communications established. I'll get into that in a little bit That's that's pretty important. Um, and and We've we have this neighborhood that we work with all the time We have a very large number of our participants in this neighborhood And what we have done is we we we did a volunteer program. We said, hey look We're going to replace your inverters your older inverters with these new smart ones But you got to agree to to participate in the program We're going to put uh energy storage systems on your houses But you got to agree to participate in the program and we had very quick update very quick update. Um In a matter of uh days, we had the entire program Filled in terms of volunteer participants But what we did is that we clustered all of these devices On to what the utility industry calls a feeder Um, which means, uh, they are basically all on the same wire Then and we can synchronize their operation and and see the impact that it has For the entire neighborhood based on what these few houses are doing Okay, and that's one of the reasons why we needed the the high speed communications because we needed to be able to synchronize The reaction of these devices based on the measurements that we were getting back from the utility and the requests that we were getting back from the utility So one of these systems, uh from a block diagram level look like, okay? Um This is a slide I stole from from early on in the program. Um, we we Basically communicated with all of the bus devices through modbus This brings me to an important point for the industry in general Uh, and and I think, uh, I'm doing some, um stuff with the Linux foundation energy portion around this as well. Um We need better standardization, right? Um I'm old enough to remember when I would have to worry if my Wi-Fi router was compatible with the very very early Wi-Fi cards that I would put into An ISA or a PCI slot into a computer, right? Um, and all of those things are are taking care of behind the scenes right now We're not at that point in the smart iot type things for renewable energy devices Progress is being made Sunspec foundation. There's a whole bunch of of sort of standards that are coming online But the compliance with these standards, um, isn't as high as I would like Um, when I did the shines program when we did all of the programming for the inverters I had two different companies one for the inverters for the solar systems one for the inverters for the batteries These are the boxes that basically take the dc energy off of the solar panels and put it onto the grid or the dc energy Out of the battery and puts it onto the ac grid Two different companies. They were both supposedly Sunspec compliant And I had to write completely different software to access the exact same functions If I just wanted to measure voltage or current or something like that, which was the exact same between the two of them I had to access different registers. I had to do all sorts of different things fast forward almost three years And I'm currently as I'm standing here in the in in this room I've got equipment where I'm doing some tests for a national lab. Again. I have two different inverter companies I still have to use different software for accessing The same functions in those two different two different devices. So there is a real challenge in sort of maintaining that code base okay I talked about this It and part of it is is that the industry is so fast moving We've started the program and products were new when we started and were end of life before we could get them into the field And if and if anybody that's listening to me right now has ever worked with utility utilities Are notoriously slow-moving organizations. Okay It was described to me best one time as saying that they're no enabled Uh, which is not typically what you see in in a startup culture. I've worked for a couple of different startups It is it is not typically what you see even in in sort of larger corporate America In the utility industry You will almost never get into trouble for saying no to a new idea Right. Um, and they have good reason to But basically what this means is is that it is hard to try new things So we were very fortunate that Austin energy wanted to try Uh, some new things with experimentation of aggregating these batteries aggregating their sponsors for all these smart inverters And and getting them to solve some real world problems But it's it's uh, one of the reasons that this this program is so rare is because Austin energy is one of the few utilities to do that the reason that they have that mindset is because If something happens and people lose power That's really bad on this same feeder that i'm talking about right now Is actually one of the communications and command centers for the state of texas for emergency response So it's it's deemed critical infrastructure um So we we we can't do anything that would threaten or jeopardize a power outage at that facility now They have backup power and things like that but but even that is not a hundred percent right? So you don't want to cause a problem The way that it was described to me once and I think this is a bit overstated But but it's it's the way that that um the mindset of a lot of utility folks are is is Changing something on the electric grid is like working on an airplane that you're flying in Right, you don't necessarily want to make a change now. I think that's a little over the top But uh, but I can see their point right so one of the things that we wanted to study was during this program is is sort of The impact of these batteries and was it always good? um The answer is most of the time sometimes it was bad um, and then um, what were the things that that based on our assumptions upfront What were the things that we knowing what we what we knew then and knowing what we know now? What would we have done different? What did we miss? Okay, so? um the the first thing Uh that I want to talk about is is the is the missed opportunity Okay, of of the standards. I've already mentioned this a little bit earlier, right? We have we have got to figure out where a system where these devices are pretty universal, right? um utilities especially do not have The staff um or the technical expertise to manage um Residential or small commercial systems Especially if they were different, right? um, and and they had to they had to maintain different code bases for different devices They have enough work managing their utility scale systems, okay? um, this is uh, this graphic is taken out of uh one of some of the california rule 21 um, uh documentation And you can see where if the utility is talking directly. That's that top and i'm pointing at my screen like that's going to help uh, the the um If utility is talking directly to the der distributed energy resource client Then there's a standard around that there's an iEEE standard If they're talking to any kind of aggregator type system The communications between the aggregator and the end resources are out of scope and not specified So so there was a lost opportunity with within this rule 21 to sort of help force this now I talked to the to the folks that are Uh, um that that did a lot of this work. I understand the reasoning behind it. Um, uh, we basically had to end the conversation with Well, we're gonna have to agree to disagree for a while. Um, but there's There needs to be something that forces that anything like even within the sunspec standard if you if you get deep enough into it There's vendor exception for almost everything you want to do. Well, that leads to Me having to write different software for every single renverter that I want to go talk to The next thing, uh, that's a missed opportunity and I've circled it here is if you look at the time scale that um You want to talk to these devices If you're actually trying to solve momentary sort of voltage control congestion management type issues on the utility's infrastructure You need to be able to talk to them Fast and you need to be able to get them to do what you need them to do Fast, okay So if you if you need them to put more power onto the grid or if you need to take them power off of the grid and store energy or if you need them to put um, um, you know Grid services, which are known as bolt bar or bar compensation reactive power compensation You need to be able to ask for that and and ask for it quickly We didn't know exactly how quickly we wanted to to do this. So we just set up a 10 second sort of Generic response time. We'd like the thing to respond within 10 seconds We actually set that thinking it was abnormally fast and that it would be hard to do and um That we wouldn't ever need it and it turned out not only would we likely have needed it Had we had good enough measurements, but it wasn't all that hard to do and was completely scalable. So so In the again, I pulled this out of the rule 21. They're talking about residential And I've got it circled there or the aggregator systems only having to respond in minutes Um or hourly day ahead, right? So now they're asking for these systems which have the capability of responding on a second by second basis And they're they're by standard sort of handicapping them um To respond slower And we found that that was Not not particularly useful, right? Like like we had some examples And I'll get into this In a bit where the the scheduling system Was reacting quite quickly and we you know, we were reacting to them within 10 seconds But we probably should have set that at a one second interval basis and it would have been more useful. Okay Now Before I go to the next slide. Um, I have worked very hard to to avoid sort of Detailed electrical engineering power distribution stuff and I have one slide that that Has this sort of detailed power distribution engineering type stuff in and I'm going to describe it with no math so you're going to have to sort of Go with me here because I'm going to use some very simplified language to describe what is actually a fairly complex issue But it is a huge missed opportunity in terms of The marketplace for distributed resources. It is one of the things that prevents That is going to start preventing Higher penetrations of distributed resources It's one of the things that that if a utility was inclined to not put more solar on that they can start to point at and say Well, we can't put more solar on this feeder because this is happening. Well They're right. There are fixes for this you probably shouldn't put more on there But it's it's a it's a missed opportunity that we see Okay, so With that warning Now I've got everybody scared. I'm like, no, there's no math. Trust me. There's no math What I've done is I've uh taken basically an oscilloscope capture of the voltage and current Going into a single house. Okay The two green lines are the voltage and that should be and this is Closest we're gonna get an app that should be A nice smooth sine wave so that up and down and up and down that should be a nice smooth sine wave okay, um, and The way houses work is you've you've actually got 240 volt power coming into but it's it's what is known as split phase So when one voltage is going up the other voltage is going down So that's why you see the two green lines doing this nice number on top of each other In a perfect world the current Would also be a nice smooth line Okay, it would be a very nice smooth Sine wave like you see the voltage doing notice that they kind of look like shark's teeth. That's there's no There's there's ways to describe that math, but but they kind of look like shark's teeth They're they're it's a sharp jagged thing and not only that but it doesn't always line up It kind of shifts in time like this and and that is, um, an interaction between Modern devices compact fluorescent lights electronic power supplies variable speed drives all the new electronics and houses That are great for energy efficiency But they are relatively poor for what is known as power quality In all of the shine's research that we did in all of these devices that we looked at This was not addressed Okay, none of them fix this current shark's tooth problem And it's going to start to become more and more of an issue as solar penetration gets higher and higher in the united states um We're seeing issues, uh in hawaii already for this. There's uh at the iEEE power engineering society convention There is all sorts of talk about this um and and years ago I gave a speech on it, um, and I had a bunch of data, uh, and I had a bunch of Distribution engineers come up to me and argue with me one of them got so mad that that They pointed a finger at me and then just walked off They couldn't they couldn't communicate why they had such a problem with what I was saying But I was like, look, we need to be talking about this and the last time I went They didn't invite me back to speak for another five years Um, but the last time I went there was like six or seven presentations on stuff like this And everybody was talking about yeah, this is an issue we're seeing and and I kind of wanted to say I told you so but um This is still a missed opportunity. This will have to be addressed It'll probably be at a lower level than um It'll probably be a lower level than then Direct device or something you'll you'll ask for in the device like like through a modbus register or TCP ip connection or something like that, but it is something that we need to watch out for Ah, I see a question came in machine learning We'll get to that in a second We will get to that in a second. I I have some some plots on that Or some some slides on that in a bit So I see the question about machine learning and the use cases for it And and somebody please remind me if I don't talk about it. I'd also like to talk about sort of some of the Some of the concerns around machine learning That a lot of utility folks currently have So if I don't get to that somebody please remind me Okay, so the next thing that you have to understand is that in the shines program one of the challenges that we had was uh integrating a whole bunch of different devices And presenting that to utility as one resource So we had infertors. We had uh energy storage systems. We had a vehicle to grid system And we had to make all of those devices and and sort of simplify them down and open ADR was a great way for us to do that but but It it didn't have everything that we wanted As a program so you can you can add extensions to it. You can add onto it But again the the that sort of takes away from the standardization of it, right? So our open ADR system We can't just turn around and connect that to a different utility's open ADR system and and accept their commands Like that handshake is not going to be universal between between us and we want to get to the point where it is, right? That 20 30.5 is is a great it like that That's hopefully long-term going to going to fix all that but that wasn't finalized And and widely accepted what we did our initial works It it came out sort of in the process and then we're like well, we probably should have done it this way But oh, well, we're we're in we're in it for what we're doing right now. It's too late now So we aggregated A car vehicle to grid and we were the first people to do vehicle to grid in the state of texas and one of the things that we discovered was that Your field research for these devices is critical Now, I don't expect anybody to study this flow chart and sort of really understand it The the fast version of this is When we plugged the car into the charger remember, I said we were doing things on a 10 second basis Every 10 seconds we were asking the car how much energy do you have in your battery? Because we had some rules that we wanted the car to be available Okay We didn't want somebody to get into a car and have have a big discharge event just happened and not be able To go anywhere in the car. So we said all right, we need to set a baseline We'll never go below a certain state of charge for this this Battery system we need to keep the battery so that you have at least 40 miles worth of range We live in a fairly dense urban section of austin 95 of the trips that we were making in in vehicles For the staff that we're going to be using the vehicle Was less than 20 miles a day. So we just doubled it and we said all right We're going to guarantee you 40 miles. So no matter what happened somebody got in the vehicle. They had the the range There are two battery systems. This is sort of a Be careful what you ask for make sure you do your field testing well before rolling a bunch of stuff out There are two battery systems in a in an electric vehicle There's the energy storage system for propulsion And then there's the 12 volt accessor 12 or 24 volt accessory electronic system Well, that battery is kept online every time you ask the car for a state of charge So remember I said we were doing that every 10 seconds We bricked the car We ran that battery down so low that when we went to get into the car to move it For some visitors that were coming to our facility, which just so happened to be the texas state public utility commission Good planning on our part. Uh, we couldn't move the car because we had killed the battery So we have to do all kinds of stuff So this is again why the standardization of how you communicate with these devices And even within the device is sort of making sure that you have good information available This particular car would give us a state of charge There are some vehicles um and some charging standards that you can't get the state of charge out of you have to get into the vehicle Telematic so now you've got a different communication system. So all of these things go to show A the opportunity is there to do all this stuff But b there's still a lot of work needing to be done to sort of understand the The ecosphere of of when you're going to do this kind of communication do this kind of work What you're going to break and what you are going to be able to get to and do Um, let's skip that. I'd say Relatively long or not relatively that's kind of a longish videos So the the two and this gets back to that machine learning question that that came in The two use cases that we had from v2g Were utility peak load reduction and real-time price dispatch now I'm going to take a quick step back and sort of describe These two for the folks that that aren't utility centric listening in Peak load reduction is massively important for every utility Is exceptionally important in texas. There are three electric grids in the united states. Okay Well, there's more than that but three main for the continental united states There are three main electric grids. Okay, there's effectively east coast west coast and texas because texas And texas because it is a relatively small grid Can't and and because of the rules of the marketplace We can't ship a lot of power out of the state and we can't get a lot of power into the state So the state of texas has to manage itself in terms of the minimums and the maximums Which is the toughest time for utilities to manage these things Um, and and the maximum is is the most critical last summer. Um, if i don't know if anybody remembers the news We hit 1,000 or 9,000 dollars a megawatt hour which works out to nine dollars a kilowatt hour electricity Meanwhile, most residential customers in the state of texas pay an average of about 10 cents a kilowatt hour So so nine dollars versus 10 cents there were utilities losing a lot of money If they had to go buy energy on that that last minute spot market So peak load reduction this is important for batteries because there are great way for utilities to save money On on that peak load Utilities in texas have to pay for their share of the transmission system based on four 15 minute periods in the summertime 15 minutes in june 15 minutes in july 15 minutes august 15 minutes in september They have to they pay millions of dollars a year based on that one hour total of time So if they have an energy storage system that they can dispatch that basically means they're not taking a bigger share of the peak It also means that they're sticking the utilities that don't have energy storage with that portion of the bill but hey And so that was one of the use cases for the car because we can get a lot of energy out of the car quickly The other one that we looked at for the car was what is known as real-time price dispatch. That's basically Buy low sell high right So you can buy cheap energy and then sell it when energy is more expensive You're literally just making money based on the fact that Once you get past the losses the efficiency of the system This is where the machine learning comes in You have to have very good predictions on a when those peak periods are coming and b When energy prices are going to go up and down so you do the right thing When we first turned on the system, we were getting commands from the the machine learning Engine that austin energy was was running And it wasn't It was it's advanced, but I wouldn't say that it's state of the art It's very good Early versions of it were a bit confusing because they had many different things sort of requesting access to the system And then there was an economic Dispatch that that picked which one to do so every once a while we'd get Really confusing things like like in the middle of the summer three o'clock in the afternoon the batteries were charging or like Uh, are you really doing the right thing? But the engine had done the right thing at that point that was actually not A bad time to charge the battery because what the the prediction had shown was that two hours from then Was going to be a really great time to discharge Okay, so they were willing to suffer a little bit and and put Slightly expensive energy into the battery because they knew two hours They were going to be able to discharge and sell that energy for a lot of money So that's when the machine learning comes in Especially for that that and making all of those predictions Which brings up I'll go to the next slide. Let's see Uh No, actually i'm gonna go two slides. I'll talk about it brings up the concern that utilities have I'll get to that so The 4cp that critical peak pricing that I was talking about When it works I don't want to get too much into the details of this chart. Um, but when it works, it's quite valuable. Okay The fixed energy storage systems that we've put on the side of folks houses those 10 kilowatt hour five kilowatt batteries Those were about seven thousand dollars each between batteries and the electronics and the installation. Okay They would pay themselves back in two years If you manage to get the 4cp events right and they have somewhere between an eight to 10 year lifespan So for a return on investment, that's not bad. Okay And that was the costs and price and payback at the beginning of the program. They've come down I just haven't rerun the numbers. My guess is this is probably more like a year and a half right now Like like instead of eight Critical peak events you would probably get paid back in six the um The catch is is that if you are basing your Your operation of your systems, there is a perception whether it's right or wrong with a lot of uh, utility folks that The machine learning agents still need Basically human eyes on them checking them and watching them and in some cases That's true, right? Like like we had one event one Peak event last summer where our batteries were being asked to charge during the peak event Which is the last thing you'd want them to do so. Uh, and what had happened was is the prediction engine had missed When that when that peak event was there was a sudden drop-off in Production weather related drop-off in production in wind power so that prices shot up Um, uh, and then the the batteries were earlier in the day And then the batteries were in the wrong position for the peak event which happened earlier than they expected And they were charging because they thought it was going to happen later So they they they just flat out missed right and in that case the batteries cost a lot of money Okay, so So that does happen and that is um, uh That is a a concern for the utility folks Um, this is just an example of uh that price dispatch that I talked about. Um We've got um Pricing on the or state of charge for the battery on the left hand side of this We've got uh price and megawatt hours. This was actually during the winter Uh end of december 2019 So winter prices don't fluctuate near in texas what summer prices do on that on that five minute spot market Where in summertime, you can see, you know at 3 a.m The the price of electricity is four or five six dollars a megawatt hour And then as I said, it'll be legally capped at nine thousand dollars a megawatt hour I have to go look up what the numbers are going to be this summer. Um, it'll be interesting to see with with uh, Covid how much that has uh shifted the marketplace in terms of pricing We know what it is shifted our users Energy usage has has been but we don't know what that's going to do for the pricing yet Um, but in this case the the system was um steady most of the day price was uh Fairly high in the afternoon. So they did some discharges then it sort of held and then prices went back up Uh early evening it discharged and in this case when you see zero percent state of charge That's what we were reporting to the utility. Uh, but remember I said we reserved space So it says zero percent state of charge, but that's what the utility thought We were lying to them. We still had say to charge left in the battery and then uh, you can see we're at night About 11 p.m Prices started to drop now in this case the engine started to charge too early Okay, so the bad state of charge this battery starts to shoot up You get to almost 100 state of charge, uh, but if they had just waited another couple of hours They would have gotten newly free electricity because the the market price at that point had dropped to like a dollar a megawatt hour So and that that's that's one of those things where the utilities folks are like Well, you know our our modeling had you know, which is not based on on true machine learning It's you know, frankly and often uh, in many cases it can be a very simple linear regression or historical average type thing Uh, our modeling has said they should have waited and they would have been right right where the machine learning engine did not get it Quite right. Okay. So those are the slides that I have. I only see one question which kind of concerns me Um, I I need more questions come on I've got to gotta have some uh, we've got a question. Do you ever have time in texas where the wholesale price of electricity goes negative? Yes we do um, and it is uh, typically we have uh, um Uh, it's typically night time. It's typically when the wind energy is quite high Uh and demand is low. So shoulder months where there's not a lot of heating or cooling Um for for folks that don't live in texas, uh, in the summertime Uh, there are times when you know, I mean even even if the outside temperature Has dropped below 80 degrees Fahrenheit There has been so much thermal energy pounded into that house that the air conditioner Struggles to keep up and will run all night long. So We have such, you know, that typically doesn't happen in the summertime But in shoulder months where it's been mild, you'll have all of this wind production They will they can only you know, they can they can shut a bunch of it off But it gets to the point where even the base load predictions are such that There's some stuff that you you can't shut off and prices can go negative Uh long-term storage so By long-term storage The question is have you had any plans for incorporating long-term storage? Are you talking sort of um like Some of the longer thermal and or you know Pumped hydro things that um, is that the kind of storage? If so That's not typically done at a residential scale and we're typically residential experts. So We basically only look at um Limited amounts of thermal storage and by that I mean pre cooling Or preheating the structure so If you know if you are if you are in a very dynamic pricing area, we've done some pricing experiments where We've looked at okay If if if the homeowner sees the volatility of the general energy market Does it actually make sense to cool their house an extra few degrees and leave it off? During leave the air conditioner off during those peak times and let the house sort of come back up That that is incredibly Depending on how much you do with that that can be very energy inefficient Right like you can only get so much of a temperature differential between the outside and inside So you're pouring a lot of energy into that air conditioning system to buy yourself a few degrees But in some cases the price swings are so high that it actually makes economic sense To do that. So we we've looked at it from that perspective I think some of our academic researchers we turn our data over to to you know over the history of our 10-year lifetime Like 2,300 researchers from 80 countries have looked at our data And I think a couple of them have also looked at that and come to sort of the same conclusion We did that depending on how dynamic your market structure is that longer term storage May make sense, but at a residential scale, we don't there's there's very few other other Entities that can that we can put energy into for a very long period of time Yeah, the the hydrogen fuel cell those are those are more of a commercial play when we haven't looked at those for For residential structures. We have looked at For the northeast we have looked at combined heat and power type systems Which you're you're using natural gas to generate electricity, but you're also capturing the heat and using that mouse, but in Texas We don't need heat with like seriously like they at one point they wanted to do some testing on one of our In our facility, you know, I was like, what am I gonna put heat? I'm sure there's some commercial entities looking at that. That's the scale. No to think about it I'll have to go up. Let's say I'll have to get back to you. I'll have to go have to go look at it I'm I'm sure some commercial entities have at least looked at it. Whether or not they've done anything with it Coming in probably should have mentioned this earlier, but For all of our edge compute that we do everything is Linux based Uh, it's about the it's about the best thing that we can we can do out there. That's not my specialty We have a data team that that does most of that work But every single thing we put out there is is like space We buy hundreds if not thousands of respirators per year any other question Well, um, if we don't have any other questions, I guess we can whoop One more. Do we use edge cloud or uh Oh, uh, do we use edge? Computing resources cloud or or do we deploy our own servers? The answer is Yes All of those things It depends wildly on what we're trying to do for our main data collection that we do that 2.5 million data points a day. We have our own servers Uh, the we looked at the the cloud computing costs for that and that was just not not reasonable for us at the time Every year we sort of reexamine that And it gets closer and closer every year, but at this point, no, we're we're doing that on our own servers. Um depending on the Uh, we're doing some field trials right now where we're doing demand response with air conditioners Um, and we are taking a much smaller subset of that data and a very high speed low latency moving it up to a cloud resource where there's a bunch of complications being done on it, uh by some researchers at at national labs and, um, uh Academic institutions and then they have control vectors that they send back down to the the edge compute resources, which again are just, uh um a raspberry pod. Um, so those are those are, you know, uh that mix Every project we I'd love to say that we have a standard framework But every project is such is different enough where we can't just apply a fixed framework We have to look at it and say, okay, how much data? What latency? Where is it going to be? How much can you? um, uh How much uh computation is going to be done on it? Um, and and then we make a decision from there Um, if a consumer's car is used for data storage for the utility assume lifetime um Yeah, so the question is is basically, you know, how does the consumer get compensated for the wear and tear on the battery? um, and There's a there's a couple of there's a couple of things first off uh, the modern variants of lithium ion batteries have such long lifetimes if their battery management systems are good at maintaining sort of Charge rates and discharge and max charge levels and sort of understanding temperature and cell balancing That we have Like like the the Nissan Leaf that we have we have Just tortured that vehicle in terms of charges and discharges um We haven't really driven it all that much but in terms of of charge cycles and discharge cycles because of the energy storage system aspect of it We've just tortured it and and the battery is still reporting a state of health of 99 percent And so we haven't actually seen it. So the the technology in fact, uh, there was a university of texas researchers That did some studies and basically they were saying that it's getting close to the point where If you charge and discharge a lithium ion battery once or twice a day um, it's lifetime Uh, depending on the exam depending on the chemistry It's lifetime doesn't really change from if you just set it in a room and let it exist Like the the fundamental chemistry breaks down so slowly at this point And the the charge and discharge is is so well managed on these things that that it would last You know somewhere between eight to ten years To a certain state of health percentage Based on charging and discharging or if you just set it there and you didn't let it get too hot It would also last eight to ten years So so in some ways that you know, there's a a You know that I think that concern in the future will be overstated But you do need to compensate the the owner of the vehicle for that energy And the the the use of that battery and that's that is That is something that the rules and regulations are just non-existent at this point So I have no good way to to show how that's going to get done It's just that the the public utility commission sort of have to to come up with those rules and regulations. They're just not there That was a good question. Thank you These are all been good questions Uh, from a perspective, I actually drive a 2011 Chevy Volt um, and uh, it's battery life is still just fine. Um, so It's Other parts of the car are not doing so well, but the battery is just fine All righty any other questions Well, I think that is about time for us. Thank you everyone for joining me. I appreciate it um My contact info is I believe at the end of the slide deck. Yes um, and uh, I'll try and participate in the the slack discussion if there's any additional discussion there as well So but feel free to reach out to me at that email address either either or