 Good morning from Stanford University. My name is Will Chu. I'm the faculty co-director of StorageX Initiative here at the Precourt Institute for Energy. And joining me today is Professor Itui, the director of the Precourt Institute. And we're delighted to have the kickoff of our spring quarter StorageX International Symposium. For those of you, our academic listeners who had, I hope you all had a good spring break and ready to resume. And we have a very exciting line above speakers, which I will overview at the very end of the seminar today. So we have already spent a number of our sessions on a very important topic, which is long duration storage. And I'm delighted that we'll be continuing the topic today. Let me just quickly review what we have covered so far. We have heard from Bob Laughlin and Andrew Ponek from Entora Energy on thermal storage. We heard from Mike Aziz at Harvard University in George Crabtree at Oregon National Lab on new chemistry for long duration storage. On the industry side, we also heard from Marco Ferrara at FormEnergy, Andrea Pettarelli at Energy Vaults, and Yor Heneman at Intervenum, who talked about a variety of energy storage, for example, through mechanical energy and highly reversible chemistry. To continue this excellent series of seminars on long duration storage, joining us today is our very own Steve Chu and Paul Albertes from the University of Maryland. So we're gonna have Paul talk about the technical economics of long duration energy storage and there could not be a better person to do this. Paul is currently the Associate Director of the Maryland Energy Innovation Institute, and he's also a faculty member in the Chemical and Biomolecular Engineering Department at the University of Maryland. He initiated the landmark long duration storage project at RPE at the Department of Energy called DAIS. And he was also additionally responsible for the IONIX program that oversaw the development of novel solid electrolytes, both of which have implications for energy storage. Paul, we're delighted to have you join us today. We're looking forward to your talk. Great, thank you so much, Will, for the introduction. It's a pleasure to be here and I really enjoyed some of the past sessions as well. So my talk today titled stationary electricity storage daily and beyond and let's jump right into it. So I have three major topics for today. First is some context and skills for stationary electricity storage. The second is some, I think some persistent challenges that commercializing new stationary storage technology faces. And I think it's important to be clear about those and understand the history of what's happened in this field, especially if you want to make some changes going forward. And then third, talk about the topic that Will mentioned in particular, long duration, which in the context of the RPE DAIS program and the paper I'll be talking about is between roughly 10 and 100 hours of storage. And I'll have some definition of motivation there and then also some techno economic analysis. Okay, so it's always good to start with the problem we're solving and what are some options or alternatives. Really the big opportunity is in decarbonization for long duration storage. There's a variety of wedge type diagrams like this for decarbonization showing how CO2 can go down over time or greenhouse gas equivalents can go down over time. I think most of these typically identify the two biggest chunks of this would be in the context for electricity decarbonization and then also fuel switching, things like electric vehicles, so oil to electricity or electrified heating, things like this. And so these are really the two biggest wedges and there's a variety of ways to address these. Obviously, scaling up nuclear, scaling up carbon capture and sequestration. Wind in solar and other renewables are a big one here. Increasing deployment of wind in solar can be done by accepting higher curtailment rates, transmission expansion, load flexibility and then finally storage, which I'll talk about in particular today. I think it's not controversial to state that pushing wind in solar along with storage as far as we can is a highly desirable option and one that should be pursued even if we don't believe in a 100% wind in solar version of our future, pushing them as far as we can. So that's why this topic of in particular long duration storage is an important one. So I wanna put some visuals up here on the scales of stationary storage. So if you look at battery storage, it's about 0.01 terawatt hours worldwide today. It's growing. This is a relatively early stationary with a mine project in California. If you look at storage in molten salts or other kinds of thermal storage, this is in particular for concentrated solar power. You can see here the two tanks that are present and where the salts are stored. This is about three times current batteries about 0.03 terawatt hours. Pump storage vastly exceeds either of those. It's about 1.6 terawatt hours worldwide. In the US alone, it's about a quarter of a terawatt hour. But if you look at all these methods which are for electricity storage in one way or another, they're really dwarfed by what we have for the existing fossil storage that underpins the current electricity system. So in the US alone, there's about 1200 terawatt hours on a primary energy basis for natural gas storage. That would supply the US grid completely for multiple months. Of course that gas is also used for other applications like space heating, but that gives you a sense of the scale of what's going on. On-site for coal plants, typically weeks to a month of coal is stored on site and nuclear might store even multiple years of fuel on site. So that's the kind of long duration fuel system behind the existing grid that we have today. This single facility actually is located very close to where I grew up in Michigan. It's one of the largest natural gas storage facilities in the United States. That facility alone is 27 terawatt hours on a primary energy basis. One other number relevant for scale is if you wanted to timeshift roughly 10 hours of the US average electricity, that would take about five terawatt hours of storage. So this is three or four times bigger than all the pump storage in the entire world today. And so if you think about a project lasting 20 years, that would be a market that might be around a quarter of a terawatt hour per year, which might seem very large. On the other hand, this is probably 10 or 15% of the projected supply of lithium ion in 2030 for automotive alone. And so it's big, but it's actually not totally inconsistent with the scales of lithium ion that I expect. I can also mention that if you add up all the storage mandates in various states in the US today, you basically get this number here, about 0.01 terawatt hours. And so this gives you kind of a quantitative context for the scales that we're talking about. Okay, so second topic I want to mention is some of the persistent challenges that people have faced historically for commercializing new stationary storage technology. So if you look at one of the big challenge areas is that there's really good alternatives for new storage technology. So natural gas figures are proven, bankable, unlimited duration, fully installed for under a dollar a watt, about 750 per kilowatt. Natural gas combines like all similar, but they're also over 60% efficient and they're a little bit more expensive, but not dramatically so, about a dollar watt fully installed. These are really good alternatives if we don't take into account, trying to achieve decarbonization. There's a variety of market structures in barriers for storage. There's a lot of discussion there. It's things are shifting in that front. Another challenge has been that the electricity industry is just a tough market. The assets that are desired are technically mature, bankable, low cost and large scale. This is not a consumer market. And so this has been a challenge. The first markets that are present typically just aren't big enough to scale. So things like backup power, micro grids, demand charge reduction, there's just not, hasn't been enough cumulative money in those first markets to make it to larger markets. And there's also another reality of stationary storage that there's dozens of physically valid approaches that also that passed like a first pass techno economic analysis. And that's probably diluted investment both in public sector and private sector and potentially contributed to us not having a storage technology for stationery that's really scaling up quickly right now. And this is a figure I like. It's amazing how fast history changes sometimes in some of these things. This is actually a slide from IHS. Incidentally, it was published five years ago almost exactly to the day. And it shows installed capacity versus technology maturity for a variety of classes of storage technologies. And you can look at the list of companies that's here and you can see some are certainly still around. Some are no longer around five years later. A variety of other things like zinc-based batteries, flow batteries, still in a discussion. And there's also some things that I think in a way have reached technological maturity. Certain software batteries, for example, our technology primarily developed for grid by NGK. They're now, if you look at as of today about 0.6 gigawatts of cumulative installation there. And there's other things that have become mature but are really not making it a significant wave on the grid. So the biggest thing that's changed the last five years is that lithium ion has really broken through into large scale grid applications. And we'll talk more about why that's happened. But I think as far as I can tell a significantly new electricity storage technology that was really specifically developed for stationary use has never been realized at above gigawatt scale. I think we need to be clear about that. There's been really dozens of companies that have tried this and it's a really hard problem. And I think if this is gonna change going forward there needs to be some strategy put in place to make that change happen because it's not gonna be particularly easy to do. Okay, so let's put some numbers on the history of lithium ion which might give us kind of a quantitative sense of what might be required to bring a new stationary storage into commercialization. And so this shows the first approximately 20 or so years of lithium ion history. Lithium ion was first commercialized in 91. Poor electronics consisted of its early years and certainly still a big market. It was probably about $75 billion of cumulative revenue for lithium ion prior to the introduction of the first, let's say iconic electric vehicle, the Model S in 2012. That's roughly how much cumulative revenue was present. And then this directly contributed to the development of the first lithium ion storage projects. There's one picture I showed earlier in California of lithium ion installation. This directly used automotive lithium ion cells from LG. And so there's a clear history here between cell technology for poor electronics, cell technology for vehicles and packed technology and then how this ended up on the grid. So it's really the grid's like the third customer in a way for lithium ion. And let's put even more numbers on this. If you look at the history of approximate cell prices for lithium ion, this figure shows on the right-hand side, the cell costs per kilowatt hour, I should say price per kilowatt hour. Over the first 15 years or so of lithium ion life, people were selling for thousands of dollars at the start, so thousands of dollars per kilowatt hour at the start of lithium ion history. This came down eventually to the hundreds, but there's actually an interesting similar story for vehicle battery packs. So even once you have cells, the kind of thing you put into the grid with the mind installation, there's a lot of probably overlap between packed technology for automotive and for grid. And so you can see are also the actual pack prices for a vehicle were also much higher initially. And so there's a lot of cumulative revenue, even early on, that helped to eventually make large format lithium ion cells and also packed technology cost effective, not only of course for automotive, but also for grid. And so just to state this with words, grid lithium ion had significant consumer first markets in both small cells, like for electronics and also in large packs and vehicles. There's also an interesting point that this is a case of kind of cross sector indirect government subsidies. So the federal EV subsidies that have been in place for some time really probably directly helped drive down prices for grid stationary storage projects. Again, because it's significant overlap between automotive, both cells and packs. And so what's going on quite a few times historically is that there's been a situation like this where a new stationary storage technology has exciting research, spin out development, demonstration, different kinds of projects have different costs for development and demonstrations. So I put a few airbars on here, but historically it's been pretty challenging to make it over this deployment wall. And so I think if we are gonna develop and commercialize a new storage technology, there's some important questions to address. There's a number of technologies out there that target range of durations. How should support be structured given that situation? Who will support demonstration projects when there's still technical risk remaining, especially if the projects really are best done at large scale, there's turbines involved or other large scale pieces of mechanical or other equipment. And then finally, what policies should be put in place to support new storage technology deployment, especially early phase. As I mentioned earlier, there's some storage mandates out there, but again, if you added up, it's basically tiny compared to the scale of the overall situation. Okay, so that kind of includes some, more high level type comments and context for this whole area. But to turn next and talk in more detail about long duration electricity storage. And this is again, this is the paper that we published. It came out a little bit over a year ago. And one point here is at RBE, we put together a FOA on this topic. And I think this paper really represents the kind of second version of thinking behind the underlying techno economics behind this problem statement. So I think would strongly recommend this paper over the FOA. I think we want to not correct exact wasn't the FOA, but kind of put the next level deeper thought behind this problem statement even after the FOA came out. Okay, so first off, of course, let's define this. What is long duration? People use that in many different ways. It's always could need to have numbers on it. So we chose for the context of this paper and the days program approximately 10 hours on the lower end, that's approximately the upper limit for shifting energy or electricity within a day. There's a lot of value in that. There's a lot of existing effort. And so we were kind of bounding that on the lower end. And then at the top end, we chose a hundred hours, you know, a little bit looser, let's say low hundreds of hours type duration. We'll see this does provide some substantial smoothing of wind or solar, but it's also not seasonal storage, which is probably a separate problem. The other comment I make is that, you know, based on our analysis, the thought was that once you get up to a hundred or a couple hundred hours, that's probably the upper limit of being able to economically do anything that involves a container, where you actually put something in the container, ship it to a site and deploy it as a stationary storage project. And so there's also that reason and we'll get into that more as we go along here as well. So here's kind of the visual version. I think Andrew Ponick showed a version of this for Cali. So in an earlier talk, this was one that we put together I don't know, almost four or five years ago in the context of days. This is for wind output in Texas. So if you look at one hour time blocks, it's all over the place between approximately zero gigawatts and 15 gigawatts. And then if you increase, just make blocks and average it within those blocks. You can see eight hours, it's still pretty close between zero and 15. If you look at 24, it starts to tighten up a little bit at 72, it's definitely gotten tighter. This is one 68. And so if we just kind of color what that looks like, you can see that, you know, roughly at 100 hours, you know, low hundreds of hours, there's substantial smoothing, even of a resource like wind. And obviously how exactly this looks will depend on whether you talk about wind or solar, a mixture of wind or solar, what exactly the location is, et cetera. But this is kind of the semi-quantitative version of how we ended up with that number in terms of its impact on enhancing wind and solar or other variable resources on the grid. And we also put together this figure. This is again, a kind of semi-quantitative figure. This shows the maximum required storage duration to meet all hours of load. So this is the hours of rated power versus annual electricity from wind and solar on a regional grid. And this is based on some excellent work from a variety of great groups doing work in this area. There's been some additional analysis published even since this came out. In the last couple of years. And so you can see here that, even 50%, 60%, probably daily is fine. If you wanna get into the 70s, 80s, 90s, that's where you're gonna need to have longer than daily or intraday storage. And then obviously there's a seasonal problem up here as well where you need to wait eight months. I should also emphasize that exactly how this looks strongly depends on how much curtailment you want to accept. Whether you have a broad transmission system and also things like grid flexibility. So this kind of shifts back and forth depending on where you're at on some of those key metrics. She also mentioned that if you look at 2019 ERCOT it's at roughly 20% and CalISO 2019 is about 30%. So this is kind of the world that a couple of particular balancing authorities are currently operating in. Okay, so let's then shift to some more detailed technical economic analysis of this problem statement and kind of think through at a more granular level some of the key ingredients and how they play into economics and technical characteristics here. I know doing math equations before 8 a.m. in the morning for folks on the in California on the West Coast is not particularly enjoyable. So I did have a few slides on this. I think it is important to kind of get into the weeds a little bit on some of the key terms. And I think we'll see some interesting dependencies and relationships that don't always necessarily jump out. So one thing that we did in our paper is to try to put economics on a full project basis. So there's no like levelized cost of storage or things here. We wanted to basically account for how are you getting paid on the left side here? This is the revenue that you make on the right hand side. This would be the various costs that would be incurred over the whole project life. And so everything here has units of dollars per kilowatt hour. If you're a person who likes dollar per kilowatt basis you can just multiply through by D and the duration in hours. And so that will give you a dollar per kilowatt basis. So you can easily jump between those two if you'd like. Okay, so if we look first at the left-hand side of this how storage projects are paid. So we have a 20 year project life. We've got a certain discount rate here of 10%. And there's really two key terms here. The first term is getting paid for energy, for charging and discharging. Getting paid more for when you discharge compared to when you charge. And so Delta E here, this is the price difference which we looked at roughly between five and 10 cents per kilowatt hour. And then also the number of cycles per year that you achieve. And then on this payment over here this is for a capacity payment. So this is a very common mechanism in electricity grids to get paid for capacity, which means you're available to provide output when it might be needed. And so this is a dollar per kilowatt year and then you can divide by duration to put this on the right basis. And there's a few kind of key points here. You can actually see directly from here. So as your duration goes up if you have say a hundred hour duration theoretically the number of full cycles that you can do per year must be going down compared to if you're doing it, let's say a 10 hour duration. And so when you multiply these things together your energy payment has to be going down. So we can directly see from this payment here as duration goes up, your revenue per kilowatt hour store goes down. And the capacity payment actually something similar happens as well. So if you're paid a certain amount per dollar per kilowatt year for capacity as your duration increases, that also goes down. And so that's both of these terms have a reduced payment as duration goes up. That's really a key dependency here. Another point is that we assume that this capacity payment is invariant with duration. There's a lot of discussion now and various market reforms being made. What duration do you need to get a capacity payment? We don't really get into that here. How this is, there might be a function at some point different places might choose different relationships, et cetera. But that's another kind of very lively and active discussion point. Okay, so look at the right hand side of this now the total cost of ownership over project life. There's two kind of really important areas here. This first one is the installed capital cost these two terms here. And we've divided this assuming that you have like a power block, like a turbine or a stack. And then separately you'd have an energy block which is where the energy is stored. And so you can see here, this is the cost per kilowatt divided by duration. And then this is the energy cost. And it's important to point out this is the theoretical energy cost which means the energy sitting in the tank without any efficiency losses. So if you're electrochemical storage, this would not take into account inefficiencies on discharge as you convert that chemical energy into electricity. And then we directly account for that efficiency right here. And so you can immediately see that this discharge efficiency is very important because it really directly affects your cost per kilowatt hour delivered, not just what you store in your tanks but what's actually delivered. The other important term here is OpEx. And really this first one is the most kind of important one here. This is how much you pay for energy that you buy electricity that you buy, that you turn into heat instead of returning and selling back onto the grid again. And so you can see here, this is the number of cycles that you do per year. This is the charging price or the price you pay to buy electricity from the grid and then here's a round trip efficiency that's present. And so this is the discounted cost that you have for turning energy into heat that's not sent back to the grid. And then some other terms here, details that we don't need to get into including replacement costs, things like this. Okay, so some interesting points here. So as we already discussed, a low discharge efficiency will impact your OpEx. Here there's a discharge efficiency in the round trip but also increases your capital cost of dollar per kilowatt hour delivered which is what you really care about. You also have to pay for, for example, heat rejection. So if you have a low discharge efficiency, let's say that you have 100 megawatts coming out of your tank and you have 70% discharge efficiency, you've got 30 megawatts you have to reject in heat. And so that also has to be accounted for that might show up in your power cost. And so this efficiency factors into multiple parts of our cost structure. Same thing for charge efficiency, there's an impact on operating costs directly through our efficiency but also again, cooling but also power conversion. So if you buy 100 megawatts off the grid you have to convert that to whatever your storage device uses. If only 70 megawatts actually end up in your storage media, you've got 30 megawatts of power conversion equipment sitting around that's not converting electricity directly into your storage media. And so it's really important to think carefully through the economics of these systems especially if you have different indoor asymmetric charge and discharge efficiencies. Okay, one other point that a lot of technology folks and don't necessarily appreciate fully is that the key equipment costs that folks often think about the most might only be roughly half or a bit more of the fully installed cost. And so every year the EIA puts out updates of how much power plants actually cost to build and install. So this is one for 2019. This is the latest one they have. This is for a combined cycle plant. This is a gigawatt scale plant. So 10, 83 megawatts. You can see here you spend about 490 million in the key mechanical equipment but in terms of your fully installed project costs it's over a billion. And so the key equipment like the turbines and other kind of key mechanical equipment is roughly half of your fully installed cost. And this reflects the fact that this is a big plant, it's like everyone this is not kind of a factory, right? This is kind of a built on site. There's a lot of pipes to fit together and land and things like this. And so that's reflected here. It's a bit different for a more containerized system like a lithium ion battery. So this shows the lithium ion battery, 50 megawatts, 200 megawatt hours to a four hour duration. You can see here the batteries themselves coming at about 40 million and the cost of the fully installed plant is about 70 million. And so here the batteries themselves are more than half but on the order of half or so of the fully installed cost. And so that's important to keep in mind is that there's a bunch of other stuff in these tables that you have to also take into account. And that's important for our thinking about our tech economics. Okay, so here's kind of the key results from this paper and it's easy to imagine capital cost and kind of talk about capital cost. And so we've isolated just the installed capital costs here and this includes both the power component and the energy components. And so they're grouped together here. And, you know, unfortunately you can't make a single cost target for this. You got to make these rainbow figures and look at a bunch of scenarios and things like that. And so let me walk you through this particular case for a minute. So here we're looking at getting paid five cents per kilowatt hour per cycle, purchasing electricity at about two and a half cents a kilowatt hour. And then you can see here we've got two capacity payments and then our axes here and we also have three different durations, 10, 50 and 100. And then our two axes of the round trip efficiency and the number of cycles that we could do per year. Okay, so if we look at this we can start on the left side here. If you imagine you've got something like with the Mayan fairly efficient, you know, 80% or 85 round trip. You can see here that you might have roughly $150 per kilowatt hour for your capital costs. If you have a higher capacity payment, it's more like 200. This is definitely not what with Mayan is installing it today in the grid but it's kind of in the realm possible for sure in the next five or 10 years. And this is a number that a lot of other storage technologies are also aiming for for kind of the intraday world. But you can see that as you go to longer durations like 50 or 100, the amount you have for capital cost is now down in kind of a five or $10 per kilowatt hour maybe 15 per kilowatt hour. So this is really a fundamentally different space than what with the Mayan is ever going to get to with the amount of active materials alone are about $40 per kilowatt hour today. And so this is not going to ever be achieved by the Mayan. You can also see that there's actually a pretty strong impact from round trip efficiency. For this particular scenario here, you know if you're not at least at 50% round trip efficiency you're really not having much funding for capital expenses. So even here you can see there's an importance to be above 50%. If you have a higher capacity payment not as much energy, it's a bit different. But again, you know, there's a strong incentive to be above 50% round trip efficiency in terms of project economics for this particular case we're looking at here. And the story is a bit different if you have a higher energy payment. I think, you know, accepting a higher cost of electricity coming out of long duration storage system is reasonable. You can see here this is kind of in the low hundreds at 10 hours and maybe in the low tens for a hundred hour duration for this case of 10 cents per kilowatt hour per cycle. But again, if you look at kind of 50 or 100 hours this is not anything that with the Mayan is going to end up hitting in the future. Okay, so once you have this total amount that you have for capital costs you can split it up into your installed capital costs or power and then your installed capital costs for your energy components. And this is a particular scenario here the first kind of rainbow diagram I showed. And you can, okay, so where would you want to be? You want to be kind of near the knee here, right? So you have a good amount of funding for your energy block and also a good amount for your power block. With the Mayan is somewhere out here might end up a little bit lower, but something like this the energy costs of pump storage hydro in case are much lower. I've also got a couple of colors on here for different kinds of round trip efficiency for something more like with the Mayan and something more like chemical storage. Okay, so you can see here if you're at for example, a hundred hours or 50 hours the cost that you're the amount of funding you have for your power block is definitely the low tens of dollars per kilowatt hour. And when you also remember this is installed and this is also theory, not delivered, right? Roughly $10 per kilowatt hour or well even below that five or so is really kind of the place that you'd want to end up to be compelling in this kind of design space. Okay, so what kind of technologies can we look at or think about here? And this is a figure put together that kind of showed at a high level what kinds of ingredients or what kinds of things we might look at. So obviously with the Mayan in this world here there's other store of technologies that could play well here as well. With the Mayan projects can range over a huge range of powers and also durations. And there's also obviously chemical storage up here. This is very compelling for long duration stores. As I mentioned earlier, natural gas is stored this way for enormous volumes and durations. The challenge with the chemical storage if it's electricity in and out the range of efficiency is probably gonna be roughly 40% or well below 30 even for depending which molecule you're making. And so this is really the challenge for chemical storage. And so the question is, is there kind of an intermediate range here that can provide long duration, have a fundamentally different cost structure than what the Mayan has been discussed and things like flow batteries or other electrochemicals technologies, mechanical systems, thermal systems, potentially including both hot and cold storage which Professor Laughlin talked about or including a very, very hot side which Andrew Ponick talked about and others of course also. These are the kinds of things that might be able to fit into this intermediate type region. What's our comment I'll make as we thought about this problem statement is that we can also imagine other kinds of storage architectures. So historically, if you think about going to longer duration, what do you do? You make a bigger tank and it takes longer to deplete whatever is inside of it. There's also a different way to think about it conceptually if you have a single power block like a turbine or electrochemical stack instead of a single big tank into which you put all of your storage medium you could actually imagine partitioning into multiple tanks that would have different characteristics. And in particular, as you go out to our longer durations that kind of segment that gets the most infrequent use has the lowest cycle life requirement. And so you can actually have kind of different attributes of tanks or storage media as you go through here. For thermal storage, for example, you might wanna increase your insulation. You might wanna store at a higher temperature or things like that in order to help to improve some of your attributes. So I think this is kind of conceptually worth at least considering whether this can be designed in a structurally different way. And then one of our comment, I think this is important as kind of a first pass check. So if you look at the capital cost of the storage medium and the containment alone, this is a good place to get started for these analysis. So here you're applying the storage media capital cost. This is what's actually stored energy. This is dollars per kilowatt hour theoretical, right? So we're not accounting for discharge efficiency versus the storage media capital cost volume metric, so dollars per liter. So diagonal lines here correspond to energy density. And you can see here, again, if we kind of remember our target of being below $10 per kilowatt hour, you can see that there's some things that fit in here, but even these are maybe a little close for the much longer durations. And then you can also translate this and just think about just buying a container that can do this. So here this is just containment capital cost versus energy density for different kinds of standard type containers. And so for example, if you pick water, that you have a 300 meter head on, the energy density there is well below 0.01 kilowatt hours per liter. And so if you come over here to this line here, you can see that if you're at 0.01 kilowatt hours per liter that you can't even really afford a shipping container, because if we need to be well below $5 per kilowatt hour, even a shipping container is too expensive at that cost. And of course, we know that water is stored in large ponds, not inside shipping containers, but this gives you a sense of what's involved. One other quick comment here is that it's important to include all the tanks that you need. And so if you've got a storage technology that's four tanks, like two on the hot side, two on the cold side, you have to include all those tanks in these kinds of calculations. And so that's another kind of point to keep in mind in the back of your head. Okay, so this is kind of final takeaways here. This long duration electricity storage topic, especially for tens of hours or low hundreds has really a very different cost structure than what the Maya, that what the Maya is not gonna get to in the future. We need to really carefully think about some of these technical economic issues, impact of efficiency, for example, on project costs, both operating and capital. And then a good first check on this is to look carefully at the energy density inclusive of all the storage media and also taking into account discharge efficiency and also the containment. That's a good first check on economic liability. And then finally, there's some unique aspects of this kind of storage, including the very different duty cycle than what you have, than the intraday that could allow some interesting new kinds of architectures that could be well suited to this kind of problem statement. And so with that, I'll wrap up. I'll thank Joe Manser who contributed to the drill paper and the day's program development. And then also Scott Litzman and Max Tubman are currently at our review work on this problem statement. Thank you. Paul, thank you very much for that wonderful overview of long duration energy storage. I certainly enjoy reading the paper when it first came out. So we have lots of questions from our attentive audience and I'm just going to try to group them thematically. So maybe let's get started on a high level question Paul. So you highlighted the diversity of approaches being pursued right now for long duration energy storage and you hinted at the potential challenge of diluting our R&D efforts. Can you speak to a bit more on your thoughts on balancing, making sure we have a portfolio of solutions but yet we spend enough time to further promising technology along? And there was also a question on your perspective on China which is considerably more focused in terms of its technology roadmap than the United States. Yeah, this is a good discussion comment. I'm not sure I'm the best. I'm not a policy maker and I think I have some personal perspectives but it's a bit of a tricky issue. Yeah, on the one hand diversity is good. On the other hand, I think as I mentioned, it's really hard to bring a new storage technology to market. And so there's kind of two key ingredients. One is you can fund more storage technology projects. This can be done by the government. It can be done by VCs or other investors including all the way to the kind of large scale demonstration type project. And that's important. The other side of it is the market pull side of it, right? And I think as I mentioned, if you add up all these storage mandates, they're tiny. This is not significant. And so I think at a high level if you kind of think about balancing those two sides probably the biggest impact would be to create more pull at the end. And I try to put a number on this a little bit like if you look at what you might. And it's not analogous because some of these systems are very different in their structure and their physical characteristics but having 10 years and tens of billions, like that's the kind of number that would be helpful to create market pull and help start sorting out for the technologies that do make it to a reasonable scale demonstration projects, which are the ones that are ready to kind of make it to the next stage, which things can a long guarantee program invest in that those kinds of discussions. So hopefully that's a little bit helpful. Yeah, we can definitely discuss this more with Steve after his talk. So diving into the details a bit more your analysis really highlights quantitatively on the importance of efficiency versus CAPEX, energy density and so forth. So that's maybe a focus on a few of those. On volumetric energy density, the plot you make, I have seen it use so many times on the cost of containers. Yet I think for many of the technologies we're looking at today, it's often mentioned that energy density isn't that important for long duration storage. And I think your analysis reveals that this is not entirely true. Can you comment a bit on where we need to be in terms of energy density and where we're at right now? How big is the gap that we're trying to address? So each, we talked about just now as a framework for doing these evaluations. So I think there's not like a single number. I think, again, the underlying idea here was can there be a containerized system that gets out to the tens or low hundreds of hours and still works? And that was kind of the impetus behind this containerization figure that we have. And so it depends a lot but a hundred watt hours per liter is probably a reasonable number to have as kind of a rough number to get much below. And that really should be a hundred watt hours per liter roughly inclusive of all the storage media on a deliberate basis. And so this is definitely lower than lithium ion. It's definitely way above pumped hydro. Obviously pumped hydro is not containerized. If you have a solution that's not containerized and this kind of number doesn't apply as much but if I had to pick kind of a rough number that's kind of a reasonable pass. But again, you gotta look at the full analysis and where it's at. In terms of the gaps, it really comes down to there's many different technologies out there and you have to know which one you're talking about and things like that. But putting forward technologies that are well below a hundred watt hours per liter especially if it's below 10 it's like you gotta really be careful about that. And the other thing is just shipping costs, right? So this stuff has to be physically moved from where it's manufactured to a site, right? It has to be staged on site. There has to be concrete pads onto which the stuff goes. The lower your energy density it also scales with pumping, right? So you have to pump more material per unit of power. And so there's a lot of scaling that happens with energy density. And I think through each one of those and how that plays into it. Yeah, but I absolutely agree and this reminds me a lot of solar once you get to a certain cost point then the weight really matters for example, for installation. And I'm just quite curious that the field so if you plot the energy density being looked at and the costs of storage I think you'll probably find a lot of those concentrated at the lower energy density side and lower cost side. And I think I just haven't seen as much on the higher energy density side lower cost side. I wonder if there's a fundamental limitation to address that part of the plot. So chemical storage is obvious when they're right. Ammonia, you could put a tank we could just find other chemical storage. That's a great option. The problem is round trip efficiency, right? And that, like it's in discussion like what are we really willing to accept in terms of round trip efficiency? And that's also kind of a multifaceted discussion. There's people out there who advocate oh, free renewables, tons of curtailment type argument. I understand that. That really then gets into like how are the project finances structured and things like that? Like who's really paying for the whole project when there's a lot of it that's not being receiving any revenue at all. But that's kind of the discussion that happens there I think and that's a good one. There are some other approaches perhaps that can get to very high energy density and low cost. It's not a huge list. And at least based on what I was saying. Absolutely. Maybe let me ask one follow-up on that and we can wrap up here and go to Steve. In your analysis, you assumed a fixed electricity cost or you also accounted for operating modes like load shedding in terms of time of use pricing. That analysis is fixed. So you buy like there it's two and a half cents a kilowatt hour, which is very aggressive. I think roughly the lowest place that industrial customer must have a special contract, I believe pays in the US today is over five cents a kilowatt hour. So even if your generation cost is really low people still have to pay transmission, right? Like half of your personal electricity bill is nothing to do with a generation, if not more. Right? So that's still in there. So you have to then get into this issue of well we do this stuff like code located with wind and solar so you don't go to the grid and pay for transmission and come back again. So I think the two and a half there we were trying to be pretty aggressive and there's scenarios of what you're not gonna even go that low. Thank you, Paul. I think much more to discuss during the panel discussion. Paul, thank you again for your wonderful presentation. So let me now go to Yi. Well, thank you, Will. Thank you Paul for the excellent talk. Let me invite Steve Chu to the stage. Let me do a very quick introduction about Steve who really does not need introduction but I'll just say a few things right here. Steve is currently a professor here at Stanford University and to many of us, I think probably all of us the audience right here, Steve was known to be very innovative and to help generating the exciting program and renewable energy and energy related research. He was in Bayer Lab before and then joining Stanford faculty meaning Nobel Prize in 97 after working in the physics for a long time, he also started the program in biophysics and later really looking into the clean energy program and about 2004 also he become the Law and Berkeley Lab director. I personally, how I got into energy has something to do with Steve when he become the Berkeley Lab directors and his speech was so encouraging and stimulating for a young person like me, I was a post-doc at Berkeley and later moved to Stanford and Steve after spent several years as Law and Berkeley Lab director, certainly really shaking up the energy program and in the whole Berkeley community later becoming the Secretary of Energy, then in Department of Energy he launched certainly multiple also exciting activity including starting RPE. So after stepping down from Secretary of Energy position and he come back to Stanford as a faculty I was fortunate to collaborate with Steve and really having a firsthand experience of brainstorming with Steve, see him how he thinks about problem and analyzing and identify the important problems and oftentimes the solutions as well and energy storage is one of the areas Steve look into and today it's very fortunate let us have Steve to share with us what's in his mind, Steve. Okay, thank you very much from that very kind introduction. Let me just begin by sharing screen. So I wasn't sure who was coming before me while meeting for my new but I didn't know about Andrew Ponson, Bob Laughlin but that's okay. So I wanna talk about energy storage and I was asked to talk about thermal storage but I decided I couldn't do that. And so I'm gonna talk about thermal storage a lot but also other forms of storage. A little note, this is a picture I love I think it was taken for me to space shuttle to the space station of a breaking dawn over Earth and the thing I really want to draw your attention to is the atmosphere is very thin. In fact, most of life on Earth is within a couple of kilometers of sea level. So we live in a really thin little space. I'm gonna talk first about energy storage in the context of what's the competition. And here's the competition. You can think going into the future as we acknowledge that the coal is gonna be phased out hopefully rapidly but natural gas will be around for decades. And so I did a little arithmetic last night and he said, well, look up one cubic foot of natural gas how many kilowatt hours at other unit of energy. And I looked up what the average natural gas storage is in the United States, averaged over five years. It's seasonal as you see on the right hand side it goes up and down. And you get a very big number. The amount of natural gas storage is, let me see if I can get rid of that. Yeah, the amount of natural gas storage is five times 10 to five gigawatt hours. Then I looked at what the United States consumes on average over a 24-hour period and it's one times 10 to four gigawatt hours. So it gives us about 48 days of natural storage. All right, let's compare this to chemical batteries. Phoebe's speaker mentioned this very briefly but here we have in the right hand side from the US Energy Information Administration. You have power and energy. Those are the two things that we care about. Power, how much power, how many watts you can deliver and also how many megawatt hours. And it turns out they're roughly equal for a lot of energy storage systems. And what we see is that chemical battery storage at night 2018 is 1.2 gigawatt hours. There's a hope. It's now about 1.8, 1.7. And there's a hope it'll get up to three. So just think of three in the near-term future versus five times 10 to the five. It's a big difference. So that's where the competition is. And you say, well, you didn't these, we're expecting EVs to really penetrate the market and one of Bloomberg New Energy Finance Projections by 2040, roughly 55% of the light vehicle sales, these are our cars, pickup trucks, things of that nature SUVs would be batteries. So let's do another calculation. Oh, by the way, I just want to, you know, that sounds very exciting, but it also tells you that 45% of the light duty vehicles will still be internal combustion engines. And they're last 15 years going to 20 years. So it doesn't look like you'll be an old EV for light duty vehicles in the first half of this century for sure. All right, but let's continue on batteries and EVs. Again, projections of what people will be doing. But if you look at all of this, and these are a spread of projections over time, ExxonMobil 2018, 2017, Bloomberg New Energy Finance is the highest and so on and so forth, let's take the highest. 600 million EVs on the road by 2040, that's a lot of batteries. And let's give these batteries a huge energy storage, 82 kilowatt hours, which is the long range Tesla. Typically the leaf is 20 kilowatt hours, for example. And what you get in terms of 600 million EVs all with the long range battery of 300 mile range, you're about 1% of natural gas storage. Okay, let's talk about natural gas storage. Well, the price go up. The price is not only not gonna go up in the near term future, it's actually gonna be more worldwide ranges. In the past, if you live in a country which had natural gas or had access to a natural gas pipeline, your natural gas costs could be very low. The Henry Hub price in the United States is ridiculously low. It's less than $3. It's hovering around $2 to $3 for million BTU. And way back when, if you look at Asia, you look at the Asian spot market, you look at Europe, it was considerably higher. It was eight to $10 per million BTU. And so what has happened is the price of shipping, liquefied natural gas has come way down. And this is a Reichstag research and analysis. They were projecting at this time, June 2019, what was happening. And the question is, and they said, well, things are gonna stabilize after that, but they didn't. And when you go from January 2019 to March, or at the beginning of April, 2020, now by the time you're in April, March of 2020, February, March, April, February, March, the world is shutting down. So all bets are off. But in January, it was not. And certainly in November, December, it was not. And so you see that the prices have come down. So natural gas is going to be a serious competitor for energy storage in the sense that when you really need the power, you can turn it on. But of course, then you need to capture the carbon and you need to sequester. Okay, where are we on non-fossil sources or non-nuclear sources of turn on energy, which in looking forward in the coming decades, you think of as just energy on demand as kind of there's some stored energy somewhere. And if you look at this, you find that 96% of energy storage worldwide is pumped storage in 2017. As already pointed out, the round trip efficiency varies. It varies what's the height difference, the higher the dam, the more efficient you get, but it varies between 70 to 85% efficient. Okay, so that's pumped storage. Thermal storage is over here and it's down here in this little sliver. And if thermal storage, it breaks out of the various things, chilled water storage, just an energy reservoir, you can chill a bunch of water if you're in a building or near building a water tank and you can use that to keep your building cooler, run your air conditioning or assist your air conditioning. The most prevalent form of storage is molten thermal storage. And that has to do with the fact that you have this solar thermal energy converters that focus sunlight and heats up molten salt and that has an advantage, it has some storage capability. Okay, if you look in the right hand column thermal storage in the pink and in the blue, the blue, most of the thermal storage stay is used for what we call renewable capacity firming. What does that mean? It means if you have a big solar thermal farm what you have is you have clouds go over that could wreak havoc with the grid system if Apache clouds go over and you get to start to get a very rapid fluctuation in the voltage or power generated. And so that thermal, this firming is used in big wind turbines, they have a kind of a built-in thing where you have a big moment of inertia in the spinning but you still have issues there and they're beginning to have localized storage just to get up rid of these rapid fluctuations. And the next biggest thing is what we call time shifting. So for example, in your solar farm you have peak generation at noon time and peak use of electricity is four hours. So this is the amount of storage we're talking about. It's not long-term storage that we just heard about. Hydroelectricity in the world automatically has long-term storage, there's all the water in the dam is backed up, China leads the world in 2013. And by 2020 they will have installed, I didn't check whether it's true but they promised several years before that in 2015 that they would add 380 gigawatts and by 2020 of hydroelectric power. Surprisingly in 2014, Japan led the world in pumped storage followed by the United States and China but things are rapidly changing. Believe it or not, Taiwan is also pretty high up there. They have a lot of mountains and that's good. But China has gotten serious with pumped storage and they will become, are the world leader today? They added about 40 gigawatts by 2020 and by 23 they will have added a total of 90 gigawatts. So there'll be far and away the world leader perhaps comparable to the other countries combined. That was pumped storage work, it's very easy. You have some water up here which has a lot of potential energy with respect to a little lower basin. You wanna generate electricity, you open a valve, you let it turn a pump turbine that has a generator attached to it and off you go. Modern day turbines are reversible so if you want to store energy, if you have excess wind or solar, you just take electricity from the generator slash motor, you pump the other way, you pump it up the hill. And so this is example of what a, this water-veined turbine looks like and that's just a variable speed generator. Why variable speed? Because you can then have some control over how much power you want. It doesn't come with a newer generators, you want variable speed because sometimes you don't want the entire amount of power but of course I wasn't showing you the size of these things. So those little things down at the bottom are people. All right, also the tubes that carry the water are quite large. There's potential for an incredible amount of pump storage in the United States. We have only using a very small fraction of that. For example, there's a project called the Eagle Mountain Project. This is an abandoned mine in Southern California in Riverside County. And they thought, wouldn't it be lovely if you can pump water up into this mountainous area and that's your upper reservoir, you can create a lower reservoir and then you have pump storage. And this is going through a long permitting process and there are many people who've resisted this, resisted the use of water. And essentially even though the state is partially sympathetic, there's a lot of resistance. I might also add when I heard about financing that if you look at the pump storage and you look at the cycles, it's different for pump storage. The dams last 50, 100 years. So we have some dams that are really seriously built in the Depression years and 40s and 50s. It's a half a century now. And if you maintain them and pay attention to them, they will be there for another bunches of decades. Of course, you replace the turbine generators and things like that, which is good because you can get more efficient turbines. But here's the thing that we don't have a thing for. If you're an investor and you want to invest in something that lasts 50, 80 years, the net present value cost of money is very different. And so it's no secret right in the United States, most of the dams built in the United States were built with public assistance during Depression years, during things where they said, this is a good thing. This is true in the Tennessee Valley Authority, all the dam systems where most of the hydropower in the United States is Pacific Northwest, the Hoover Dam and other places that the government stepped in and said, we're willing to fund this long-term infrastructure project. This is a report about the cost of pump storage. If you have an existing dam and want to put a small holding pond below the dam, pump storage is very expensive. Other people looking in novel ways of pump storage because they say, well, you don't have the mountainous areas all around the world, it's only a small part of the world. And Phil Lubin, it's UC Santa Barbara is thinking, well, how about using the big vertical depth of an ocean to do pump storage? Remember, pump storage is really converting potential energy to kinetic energy, to kinetic energy to generate the electricity or taking electricity going to kinetic energy, which is very efficient. And then using that kinetic energy to pump it back up. And so Phil's idea is you can put little canisters off the coast where you don't have to go far out in many regions, Hawaiian Island, Japan, off the coast of Great Britain, other places where you can get into deep water pretty quickly. And in these little canisters, you have these little tubes and you're essentially pumping air into the tubes to displace water. And as you displace the water, you're actually working on the vertical height of the sea level to the bottom where these things are stored. And so the good news is it takes advantage of a lot of deep sea exploration and oil and gas. Finally, pump storage, you can take a salt dump, you hollow it out and you can, and this is how you hollow it out, you just pump and water to hollow it out and you can use to pump storage. We have around the world two pilot, what I would call pilot, medium large scale compressed air storage, but it hasn't really taken off. And one of the reasons as you start to pump the compressed air into this, it warms up. As you warm it up, it's harder to compress because the pressure is also proportional to temperature. And then when you want to use the compressed gas and expand, it cools down. And so step one is you have a heat exchanger. As you're compressing it, the heat goes into some stored pool and so you can get a lot more higher density, higher pressure stuff at a cooler temperature. And when you let it expand, you transfer the heat back into the expanding air and that improves the efficiency. That's adiabatic compression, but it turns out if you really think about what you want is isothermal compression. And what you mean by isothermal compression is that if you have a really good heat exchanger, i.e. the earth is your heat reservoir as you're pumping in the air and it's getting hotter, it throws it into the environment. Seems very wasteful. But remember, you're banking the fact that you're gonna get good heat exchange cheaply. And so when you have this compressed air down below and you let it expand, it's taking the heat from the surrounding area and warming it back up. And in principle, isothermal compressed air storage can be 100% efficient minus the frictional losses of motors and things like that. But we already know from pumped hydro that those frictional losses and the fine efficiency of the motors is not bad, but you're getting roughly 80% efficient. This isothermal compressed air storage is yet to be demonstrated but people are beginning to talk about it. All right, utility scale thermal storage. The driving idea is what Carnot taught us back in the 1800s, that if you want an efficient engine and you're taking energy from a hot source, so this Q sub H is the heat energy that you're taking out. And then, of course, he tells us you have to throw away the heat waste heat at some lower energy, T cold. And so he found that the mechanical work you can actually get of the temperature difference of T hot and T cold, that thermodynamic efficiency is given by T hot minus T cold over T hot. All right, so that's the driver for all thermal heat engines. One of the most efficient heat engines are the so-called gas turbine engines. The gas turbine engine works in the following way. It takes air, compresses it, you inject gas in this section in here, you heat it up, the rapidly expanding gas is then taken out and it spends this turbine. Why all these stages? It's actually quite clever because when you go into a series of compression stages, a series of expansion stages, you can get something very close to the limit of infinite number of stages and no heat loss out the size, you can get to 100% adiabatic compression and expansion, which is great, that's what you want. These turbines working extremely high temperature and the most efficient thermal plants we have today have a Brayton cycle, that's what this turbine is and a Rankin cycle, which is a more traditional heat recovery, spinocene turbine, condensers the water, make it go back. A little digression in turbines, I can't help this. This is a modern jet engine, this is a turboprop engine, which actually came before the fully jet engines in commercial flight. And then finally, the engine we have today, which is a turbofan, which is kind of a mixture. It's really a turboprop, but the propellers are now big and they're inside the engine cowling. And about 80% of the propulsion of a modern turbofan engine used widely commercially is this. It's essentially a propeller, you use the jet engine, you get this isocentric compression and expansion, you inject the fuel in here and out goes the waste heat. These things are amazing. When you're taking this and you're compressing air, you're heating it up. And just as you have to overcome this compressed air storage in these underground soil caverns, what you want to do is you don't want, you want to minimize the heat, but mostly as you're heating this up, you're exposing these blades and these blades to extremely high temperatures. And Carnot says, you got to work at the highest temperatures possible. And so if you look at these things and air bleeds and various things of compressors and in the turbine stage and you look at the fan blades, it's kind of crazy. The fan blades are actually cooled. There's cooling air that's going here and they've drilled little laser holes in these fan blades and the air goes in here and it's pulled out with centrifugal forces and it creates a little laminar sheet of coal there that protects the fan blade from the very, very hot gases. So here's a blow up of these blades. We first went to single crystal blades of metal that can be laser drilled with holes to create that laminar flow. And as the air goes through, you keep the fine base cold. How well does it work? Well, scarily well. These are the softening temperatures of the best alloys you can come up with. And the jet engines are running a 1600 C where you wouldn't dare go above 1000 C if you didn't have fan blade cooling. All right, so let's go back to these wonderful Brayton turbines. The idea here is if you have heat in, instead of injecting jet fuel, you put heat in and you heat up the air, you spin the turbine, it goes outward, the turbine after the hot gas expands, you, it's colder, you throw the heat out and heat exchanger, just like Carnot-Tolis and you then recompress it, heat it up and off you go. If you wanted to think about this in terms of a temperature entropy diagram, going from one to two you're compressing and so these compressors can be very, very efficient as I indicated. So that means this is a perfect compressor. There's no change in entropy in getting up to this point. And then over here you're putting heat in and as you go from two to three in this Brayton cycle you're heating up three to four. Again, these turbines are terrific because they've got many, many stages. So you, the gas expands and you go down again, no change in entropy, no change in entropy is very good. And then finally in the heat exchanger you're lowering the temperature and you're ejecting the heat. So during this whole process here you're ejecting cold heat, that's wasted heat. And during this whole process here you're putting in heat. What's wrong with this? Well, Carnot says the efficiency is T hot over T cold but that means if you're putting in heat, remember in this textbook example, now textbook example you have a piston on a hot reservoir and you're putting in heat all at T hot whereas over here in real life engines you're going from two to three you're putting heat during this whole thing. And so a lot of the heat you're putting in is a lower temperature over here. So your T hot over T cold is not quite as good as you want. Over here it gets pretty good, all right. So what have engineers done? Well, they said, well, let's run this compressor turbine heat but we're gonna add a little, oops, sorry. We're gonna add a little. So here's the heat in the jet fuel or the thermal source of heat, expanding turbine. Normally you just run it backwards but now through a heat exchanger but now you take some of the heat in an intermediate step and you put it through a heat exchanger, okay. And so what you have here is when you go from one to two to five to three around this cycle you're going up perfect compression, you're sticking in heat but when you get to this very high temperature and it goes down, you begin to slurp off some of the heat in this intermediate region. We're not gonna try to get all this heat to dump it as slow as we can and we put it back in here. So there's a heat exchange mechanism that takes the heat over here and sticks it into here. What does that do? It means that you're throwing heat out at a lower temperature. You used to be throwing heat all over this cycle but now you're confined into a lower temperature down here instead of averaging over this full swing and the heat over here gets shuttled over here and so that heat takes you to two to five which is just kind of taking the hot and putting it over here and now you're putting in the input heat H in at this higher point. So you're getting close to what you really want. Put heat energy in at the highest temperature, take heat energy at the lowest and you don't want to have these large swings. Well, if this works you might as well do the whole fit and throw more turbines at it. So you're kind of interkeeling, reheating, regeneration and in so doing what you have is now in these stages if you work your way through it you can leave as an exercise to the students. You're dumping heat out going from 10 to one whereas 10 back to one, you're dumping heat out. You're dumping heat out over here and this is now in the temperature entry diagram. You're really down here low. All this other stuff has been regenerated due to these heat exchangers and now you're putting in heat here. And so it gets more and more efficient. All right, now let me talk about Carnot batteries. I just learned today that you had Bob Luffin give a talk and so the basic idea of a Carnot battery is you have some excess electricity. You can use it to heat up the ground but wait a minute if you have some excess electricity or whatever, why not store it in a heat reservoir? The first order of business is storing heat reservoirs. You use the electricity directly into a hot reservoir but there's something very important that's running to my entire talk and that is the conversion of mechanical energy to generate electricity, rotational motion and going and electric motors is a very, very efficient process. And so you're immediately starting to think you don't wanna take this electrical energy, stick it into a resistor and heat it up. This is a very high entropy law system and so it's better to use heat pumps to take energy from cold and hot reservoirs because again, the electrical motor, the pump refrigerator of the heat pump is more efficient. So this is the fundamental idea of a Carnot battery, heat battery and the discharge is some kind of cycle. People are naturally drawn to the Brayton cycle because those Brayton turbines have become very, very efficient rather than a phase transition cycle like a Rankine cycle and then the heat reservoir there's all sorts of ones. Siemens looks at the storing heat in a bunch of cheap hot rocks. And so they're looking at this, high power I'm gonna get to in a moment and then MOLTA I'll get to in a moment which is a new company that's spinning off of Bob Loffin's idea. His idea is fundamentally the following, you've got hot reservoirs, you've got cold reservoirs and so there are two essential ideas, new ideas that he's introduced. One is that you wanna go to as high a temperature as possible but then as you go higher and higher temperatures you run into materials problems. And so he said, well, you know, with existing materials you gotta, and again, you're not gonna do all the tricks quite all the tricks that I was just talking about with the current turbines but you can have your heat stored, not actually storing hot molten stuff or cold stuff is really easy. You can, if anyone's played with hot stuff you know that with five or 10 centimeters of firebreak you can be red hot to orange hot on one side and you can hold your hand on the other side. So, and then it scales beautifully because the heat loss goes as the area of the container and the heat energy scales as a vibe. So it's good for, great for utility scale storage. So in Loffin's scheme he had two cold temperatures, two hot temperatures and he had a brightened turbine where again the magic, you really have to pay attention to it in the scheme is how good are these heat exchangers? And this cold temperature for example if you want to run the thing you have a hot temperature and you use it to spin the turbine otherwise you use electricity comes in to shuttle heat back and forth. So I'm not gonna walk through this but the heart of it is a brightened cycle turbine for temperature reservoirs, not two. And if everything were perfect, if the compressor were perfectly adiabatic and the heat exchangers are very, very good. This whole thing becomes thermodynamically 100% efficient. And so he says, well, maybe if in realistic numbers maybe you can get 70% efficiency by maximizing the temperature difference between the cold and hot reservoirs. Again, with a clever tweaking of the Carnot cycle. And so Malta picked this up and they, as in all slide decks looking for money they show you how important this is when we went from the stone age to the metal age to the iron age to fossil fuel age and of course the storage age, in any case. And in their slide deck they say, well, there's pumped hydro over here. This still remains the biggest long-term storage and highest power you can have. Hoover Dam, for example, can produce a couple of gigawatts of energy. So it's actually out here. And they think that Carnot batteries would be in this size in here. The Lithium-Ion batteries are actually, they've made Lithium-Ion batteries that are in this range already. But again, it's the cost, it's the problem with these batteries. I'm going to talk about something. This is I cribbed from Wikipedia and this is cryogenic energy storage. So again, we're thinking of energy differences. And so you can think now of using mechanical, this really good mechanical electricity to compression is a good thing in compressed air storage and in brightened turbines and everything. So this thing comes back over and over again. And this time, can you use electrical energy to cool something down, compress it? And now that's you put energy, work into this and you've got something cold and you've got a temperature difference, which Carnot says then you can do some mechanical work and then you bring it back up. And so if you look at this Wikipedia article, it says in isolation, this is pretty crummy. It's only 75% efficient. But if you use 50% of you use low-grade energy store and if you are near something like a power plant that's throwing energy out, they try to throw energy on not at 100C but maybe 60 or 70C. But if you have something there, you can boost a round-trip efficiency to 70%. And usually when people are thinking of any long-term energy storage and you look at the economics of that, once you go below 50%, you begin to lose interest as Albergus told us. So this is the high power thing. It's again, power in, you compress it as when you use that, you store the heat in something over here and then you refrigerate. You have high-grade heat storage over here. This is a cryogenic heat storage. You expand it. So it's again, more of the same of compression, storage, heat transfer, these other things. I have a little movie to explain in more detail. It's not my movie, it's Hightower's movie. Let's see if it works. Air is first cleaned and dried and then refrigerated to a series of compression and expansion stages until the air liquidifies. This process is based on the Claude cycle, which is over one hundred years old. The liquefied air is then stored in insulated tanks for low pressure. These tanks are really available from the industrial gases industry or a very large scale, the LNG sector. When power is required, liquid air is drawn from the tanks, pumped to high pressure, heated and expanded. The resulting high-pressure gas is then used to drive expansion turbine generators. No fuel is burnt in the process, resulting in the exhaust producing clean, dry air. Waste cold from the stage is captured by the cold storm. This is later recycled to enhance liquid-faction efficiency. And in a similar way, heat generated from compression during recharge is captured by the thermal storm. So that just took you through this cycle. It's easier to explain by videos and movies, but it's again, a slightly different combination. And unbelievably, this looks, is this seems to be being looked at seriously by a number of industrial farms and so they're looking at the high-tower steam. All right, the last thing I want to talk about is thermal photovoltaic power conversion. The idea is pretty simple. What you have is you have something, you have something very hot and a thermal emitter, and then you have a PV cell surrounding it. This is taken from Eli LaBona, which Harry Atwater and colleagues. And the idea is that this PV says, well, I've got a thermal emitter, it gives me black-body radiation, but if the thermal emitter is very, very hot, you're moving that black body near into the infrared. So much so in the infrared that you can start to take advantage of narrow band gap semiconductor materials. And so you design it so that when the light goes and if the light has high enough energy, so it actually excites an electron from the downspan to the conduction band, you can get your separated charges and you can slip off the electrons so you get your current, your photo current. But what about the infrared light? Well, the infrared light that isn't absorbed is below the band gap. They said, let's put a reflector here. And so this infrared light isn't absorbed by an ampere, it goes back and keeps the thermal emitter hot. And so in this paper they showed in this work as you go harder and harder and harder, you're getting pretty close to where you think the theoretical maximum will be, 28, 29% efficient, which is pretty good. Very recently, this came from an article that was posted in, it's a news article, I should say that was posted in February of 2021 where you have energy storage tanks and the idea is really the same. And now it's molten silicon tanks over here and the molten silicon is then piped out. You, where do you put the heat input? Well, the heat input can be home from various things, nuclear, natural gas, coal, well, not really because you wanna get rid of the natural gas and the coal and a lot of people wanna get rid of the nuclear. So it's really winded solar. So anyway, so that you store the heat in here, the idea is you then when you want to do it, you pump it back into these little modules, multi-junction modules, they get to be very, very hot. And you have in these modules this integrated solar cell which then says, okay, it only absorbs the light. It's very much the same idea. You just convert it with maybe 30, 40% efficiency. Okay, and there off you go. So these are some of the ideas that people are looking with because you can't get, I think people should really look at compressed air storage. You should maximize all the pump hydro storage you can, wherever you have a dam that's high enough, look to make a small 1% holding pond below their dam and that would be a significant energy storage. And you've already got the electrical infrastructure, the pump infrastructure. But these are some of the ideas that are around in looking at energy storage. And with that, I don't know how long I went, but I'll stop there. Yeah, Steve, thank you. This is a really good overview on many, many possibilities right there in technology. This is a question I have and also I saw from the audience also have giving so many different type of technologies right there. So how could one go about to analyze the use case for each of these technologies? Which one will likely to be dominating, depending on the use case and your location? How to think about this problem? Yeah. Okay, so there's other things that's a purely technical question, but when you think of siding wind farms near where people can see them like in Cape Cod, as you think of that, then all of a sudden you run into long years to decades delay, which costs a ton of money. And so that's the other thing that has to be folded in. Why do I'm talking about that? It's because some of these other things which are much more compact and have a smaller footprint will not have as much public resistance. The other thing is in terms of use case, you have to decide in the case of heat storage, it scales very well for utility scale storage. So you try to prove a principle with smaller stuff, which of course is not as efficient, but that's okay. You understand that. And it really boils down to how, when I look at these things in more detail, it boils down to something very mundane. How good is the heat exchange? Very good heat exchange means big heat exchanger, which costs money. And so everything has to do with this balance between CAPEX, the OPEX, the fuel you're gonna say is hey, we're gonna assume you're gonna get the energy from surplus energy. It's energy storage, right? So the only cost there is the infrastructure to bring the energy in and to port the energy back out. And so that makes thermal storage more attractive because if you look at the coal plants around the world, which are being phased out, you've got a footprint. You've got electrical infrastructure there and you generally have enough area around there. If you look at the size of the coal mountains and what you would need for a thermal energy storage and coal hot tanks, you've got some of the real estate there. So that has something going for it. And so these are some of the factors that you all have to think about. But in the end, it's how reliable is it gonna be? How well can you drive up the temperature differences? And again, always thinking about turning things. It's, there's a term that thermonics people use instead of entropy, they use exergy. But it really boils down to, if you wanna take conversion, think of, you know, you want mechanical heat. You've got temperature differences, you don't. So whatever you do, you wanna minimize that loss of entropy in any way you can. And that's why these, for example, the great turbines are so wonderful. Those many, many families take things in steps and they adequately allow you to compress and expand. Yeah, yeah. So there's a few more questions. I think also related to Paul, Albertus, maybe let me bring back Paul and also Will. Let's just go into the panel discussion. I think they're so correlated with all these questions. One question I have for both of you, certainly Will, feel free to chime in as well. And I look at the choices. So I come back to this question. There's so many technology choices right there. Paul, you have this really excellent analysis. What's the course we should think about? You know, if I look at 100 hours kind of storage, I think you give the number right there. We really want to get down to somewhere per kilowatt hour, you know, $10 type of range, right? $10, $20 around that. Certainly there's some assumption right there. And now looking at the technology choices right there and also the use case. There's kind of couple of hours storage, four hours that's lithium-ion cover really well. Then you go up to 10, you go up to a few days. And if I ask the question, would somebody want to invest for different use case in terms of time, duration, different technology? But could you think about if just one technology can do the job, then you actually, you know, your investment in return is much better, right? Because your use case is you want to have 100 hours. By the same time there's 100 hours if you already invested the storage, you want it to be able to handle, you know, several hours. So how do we think about this complex situation? Like there's a different timescale requirement. This technology have different cost profile. And also, and this use case is not like independent. They're kind of mixed together in terms of hours. So how do we think about this? I don't know who wants to take it. I thought you were addressing it to Paul. Well, maybe Paul can take it first and then Steve can also come as well. Good question. So I think most of what's going on out there today is addressing this like, let's say up to 10, 20 hour duration. So a lot of the thermal storage technology, flow batteries, mechanical type storage, et cetera. Most of this stuff is not going after the 100 hour problem statement. There are a couple that are trying to go after this really long duration problem statement like over 50. But that's by far not the majority of what's going on. Oftentimes when people talk about long duration, I'm talking about eight or 10. And to me, I think there are a variety of viable approaches. Professor Chute has talked about a number of them that could fit into that kind of single day type world. Also on this issue, what also might happen is that because lithium ion is ramping up so significantly for automotive that, you know, if someone made a, well, I think what could end up happening is that we end up with a lot of lithium ion in the grid and that's gonna take a big chunk of that intraday like 10 or 12. And then once that market of daily time shift is largely satisfied, then the longer duration, which of course can do shorter durations, won't necessarily have that market around. And so, and I think Andrew Panic talked about this a little bit in his discussion also, is there a separate market for like 10 to 100 or this longer duration type? And what would that look like? What kind of attributes might be there? So I think there is kind of conceptually and in terms of what the market looks like, there is gonna be a lot of this intraday type work and lithium ion and other technologies for the coming decade or 15 years or beyond. And so what position does that put this longer duration technologies into? And I think that's an important issue to think through. In the free market philosophy of energy storage and energy in general, first comment is that most of the energy and energy markets are now being sold as next day energy. And as we learned from Texas, not enough thought has been given to building in the resiliency of the system. The idea being that if things get really expensive, then people with energy capacity, they can turn up even though these so called peaker plants aren't used that much during the times when you really need the energy, they can charge a lot of money for the energy. That works up to a point because when your energy rates go up by a factor of 1,000, people get very upset. And not only that, in that economy, those people really don't care if they run out. And what's a few days of blackout in the overall scheme of things, while we learn a few days of blackout can cost lots. So this is, again, we have to rethink this. How much of the premium it used to be in the old days where you had a utility company that's vertically integrated, you had to be responsible for it and the utility commission said, here's a lot of money, you can build capital investments, your rates will be included in the cost of money for those capital investments, this is all great stuff. And so they took a longer term view. That was, we go more to this market economy. You're building into the system actually less resiliency, but so does going to more wind the thermal as well. So you've got these things going towards kind of a more unstable system. The gas peaker people are saying, you know, if it's only beyond five or 10% of the time, but I can charge a lot of money for that, I'm still willing to invest in it. And then the question is, the battery people might say that, but what's the ace in the card? Because you're still in this one day stuff. And when you have a five day event and there have been five days or even longer when wind has stopped blowing in regions in the United States or you have these cold hands, then what happens is then you need some really longer storage. And that's why I spent so much time talking about natural gas. Yeah. Right, we do not have something at the present time that will get us to the five day, 10 day storage time, which is kind of magical because once you're in that longer range, as Paul said in his article, you can get to a very high fraction intermittent renewables. So I mean, speaking of energy resilience, I asked a lot more questions, maybe we will also have questions to ask coming to the energy resilience and it's looking into natural gas. This is amazing, right? Steve, Paul, your estimation is roughly the same. Like 50 days, two months of the whole nation's reserve is right there. Would it be crazy to think about a national reserve for electricity kind of stuff right there? This also sort of require the breeze to be interconnected somehow. Then once, I think it's probably become a lot more robust if they can all be interconnected. Then you geographically, solar and wind, this fluctuation in different region of the nation, you kind of balance out that on one hand. The second one is once you have a number of reserve location, electricity storage, you can kind of support each other a lot more. Would that be crazy to think about national reserve of electricity storage? No, in fact, it's necessary. If you're going to get anywhere above 50%, then 75, 80% renewables, other than hydro, you need that. Again, there's resistance. People in areas where there's cheap electricity like Pacific Northwest actually don't really like cheap electricity leaving their area. Even though it's much more, it's much cheaper. When I was Secretary of Energy, I tried to get going, even energy balancing as a prelude to this. And all sorts of people hit Congress to say, don't let the secretary do this. They do not want long distance transmission. And this actually started when I was LBL director. I was in an energy study called America's Energy Future. And I was talking about a DC high voltage backbone grid that can use four time zones. And different weather patterns. We're very blessed with different weather patterns. So the wind can be low in one region, not in another. Europe wind is not like that. It's kind of the same. So we actually had it good. And I said, this would be great. All the wind from North Dakota to Texas, it's great. It's cheap land. The people love it. They make extra money, but you've got to get to the coast where the people are. But no one, there was huge resistance to building long distance high voltage transmission lines. And the average time of siding was 11 years. And I said, we got to get this down to three or four years. But I couldn't make it go in. Secretary of Interior, Ken Salazar, was very sympathetic here. But he could make the people in the fish and wildlife game part of his section go along with this. But they don't want high voltage transmission lines where they're hunting and fishing. So it's crazy. Now, China doesn't have this much resistance for something like this. So they have the best high voltage long distance transmission lines in the world. We need some of the top down. Can't be able to do this. China is very strong. I prefer still to live in a democracy, by the way. One interesting point here to look at is if you look at this is often under the radar, the natural gas transmission system is being significantly expanded even today. And the EIA publishes on this all the time. So they're like 36 inch diameter pipes. They're buried under the ground. People don't see them. And a single line can carry many kilowatts of chemical energy. People don't even notice what's happening. Or it's not nearly as widely understood. So I think that's the issue of do we need to have a nationwide or some other kind of large scale storage plus transmission. Chemical storage, hydrogen, which, of course, Europe is much more focused on, or even synthetic natural gas. You pay the efficiency penalty. And that's a really important problem. Or it might be unavoidable, especially if you go through combustion at the end. But that is what it looks like. It's like what we have today. And people don't care about putting new chemical transmission lines in. They don't even know they're there. People on the side of their backyards didn't realize a big 36 inch natural gas transition place. Yeah, maybe transition line, the electricity one, can be also buried under the ground. No, that's different because it's limited by the strength of the dielectrics. And so where you can get 1.1 megavolt DC lines in the air, there's bare wires. The highest voltage DC lines are 500 kilovolts. There's experiment, you would say, 200 kilovolts. But it's the losses v squared over r. So it's a big deal. The gas pipelines, people have been talking about hydrogen. You can't use the gas pipelines as a steel embrittlement. So you've got to lay in polymer pipes. But you can maybe use the right of way, which is part of the battle, and to install the hydrogen. But hydrogen has its own issues. The inefficiencies, as Paul mentioned. But we're going to have hydrogen and some pipelines in hydrogen for sure. Because we need all the things. We don't seem to see any magic savior coming in at us. And I still think we should maximize from nitro. Yeah, well, let me build off on this a little bit more. Steve, I know you talked about some of the political challenges with improving transmission. But if we only look at it from a technical and economic perspective, we are putting a huge amount of resources for energy storage, starting with short duration storage for lithium-ion battery. If we put that resource into transmission and benchmark that, how does the economics play out? So if I can cross four time zones, then I don't need to have storage for the four hours. And does the economics is it favorable compared to deploying local stationary storage? Well, actually, I didn't really do any serious economic analysis. Paul, have you looked at that? Not serious. I think it's a non-starter, practically. So I haven't looked at that in detail, like how much the cost per lot per mile, et cetera. The reason Paul and I think it's a non-starter is the resistance. I mean, look at the Northeast of the United States. It's got some of the highest election bills outside of Western and Alaska and Hawaii. Maine through New York, New Jersey. Really high energy costs. Canada's got tons of hydro they love to sell. Cheap, clean hydro. And the people who just bought gas generators say, no, we strand in acid. So what did they do? They got people say, we don't want to see high voltage transmission lines in Vermont and New Hampshire and other places. So then you have to road away. But for the first decade or two, that's been the thing. Now they're beginning to think of transmission lines into Canada because Canada's swimming in hydro in Quebec. And it makes no sense, right? And there was no issue of energy security. The Canadians are more trustworthy than we are. So I think we have to start thinking of these issues and transmission has to be part of it. One recent example of this is actually a former RP colleague Tim Adele, who's now at the engine MIT, started a company called Deer, which is focused on using existing transmission corridors and increasing the amount of transmission, the amount of power you can transmit on essentially the existing infrastructure. That type of thing, yes. I think other stuff just, again, this is not technical. This hasn't happened historically. It's really hard to do that. Yeah. I've been advising Germany over the last half decade or so on these things and they have a transmission line issue. This starts putting in DC lines. On a tower, it has six lines put in two DC lines. You can have three times as much power generated from the same wires, two wires. And because you can go to a higher voltage, you don't have the losses due to radiation and inductive coupling to the ground, all that stuff. Thank you, Steve. Let me maybe also ask a question about scaling up and the cost learning curve. Both of you touch upon this a great deal. And I couldn't help to make the comparison between the cost learning curve of lithium-ion batteries versus this more larger scale chemical energy storage or whatever energy storage on the size of a power plant or a chemical plant. It's quite natural to see the difference. Lithium-ion batteries, we're making billions of them. We're building many factories a year. We're learning very quickly. And so the iteration time is short. For the power plant slash chemical plant style energy storage, the learning curve is slower because the project time is much longer. Do you see this as a fundamental difference between a scale-up that's done by scaling manufacturing of small devices versus a scale-up by making something large and enormous? Could you both maybe comment and contrast these two fundamentally different cost learners? Let me start. That's a great observation and comment and question. And what I see is the only way you can do this is to start to stamp out turbines and things that are standard thing at a much higher volume in a single factory. The pictures I showed you of hydro were one-offs. Okay, you can't do that. Just as the nuclear reactors are built on site with expense costs, you want to go to a small modular reactor where you stamp a amount of factory by the thousands and you have higher quality control, but even the small modules, they're thinking of doing it, but you've still got the rest of the onsite stuff. So the more you can get a deliverable that's made in a factory and the whole thing goes there, the better you are. So I started thinking about this and started saying, can you standardize turbines? Can you standardize degenerates? Can you standardize them? And you go to these two or three sizes and start mass producing like crazy. And then that's the only way you can, if you start doing one-off turbines, this is crazy. And so that's your only hope. Now you're never going to get to the car manufacturing, cell phone rapid turnover because these things are going to last a half a century or more. Well, the motor generator part maybe 30 years, okay? But so you go to mass production in high volume. Thank you, Steve. And maybe also the aerospace industry might be a sort of a middle ground for comparison as well in terms of medium volume manufacturing. Paul, any thoughts? Yeah, there is a fundamental difference there. And I kind of try to provide a little bit of numbers in the history of lithium-ion there and what it took. One, obviously lithium-ion, something like this, this is a new chemistry, right? And it sounds like it's just one thing, but there was a lot of different pieces and new development. It's something truly new, right? Some of the thermal storage technologies, they're still using turbines. As Professor Chu mentioned, there's a whole bunch of different configurations and things like this, but there is a bit of a shelf there that there's more stuff on compared to something like lithium-ion. So I think that makes them a bit more comparable, but being able to start small, get to try things out, do the learning curve, that's very important. One other comment I'll make is who's buying it, right? Where's the poll? So we can do all this stuff and talk about pushing stuff out there, but someone's got to buy it. There's still that issue too, is that who's going to sign up for 10 years and tens of billions of dollars of purchasing grid storage products? If that is there and secure, that does change a lot of investors' calculations. And some of these things kind of happen automatically, or not automatically, but they happen because people see that there's a market that they can go into. And so that would create a lot of activity that I think it's going to be hard, even if we have ideas like standardization and stuff on the technology side, without that poll, it's really going to be a challenge to get things to happen. Absolutely, Paul. Yi, would you like to wrap us up? We're at the end of the hour. Yeah, I think you should go ahead and then also advertising for the next event as well, yeah. Sure, well, Stephen Paul, do you want to sort of share a one-minute outlook? We talked a lot about challenges and opportunities. So maybe you can say a couple of words on what is to come. Well, I'll begin. I think the energy storage and transmission and the integration are the key issues facing us now. The learning curves of wind and solar are great. They can continue. In the transmission side, we need, I'd love to have an efficient diamond transistor. So you can get DC and go up and down and tap off in many places instead of just one stop. There may be some a few more tricks. So if you think of the evolution of the Brayton turbine and how it got better and better and better, it's kind of amazing what they're doing, but it was different because it was in for a commercial airplane where a few percent made a lot. That was the market pull. And I don't think we have, as Paul said, that market pull in some of this stuff yet. Yeah, I think, Akio, this is an important topic. And I really do think that if we think about the suite of things that can be done to do decarbonization, that wind-sower storage, that's really important. And having read thousands, probably, proposals as an RP program, there's a lot of good ideas out there. We've seen a lot of companies in the symposium series and others who have not presented here. There's a lot of excited technologists who are working on this problem. And I think I just echo again something I tried to comment and Professor Chu also mentioned is that where's the market pull on this? And the mandates are just, state mandates are tiny. And so I think that is a kind of missing piece of this as much as we love, and I love talking about storage science and technology that there's still something at the other side of this that's still, I think, missing. And it's worth thinking carefully about and seeing what can be done there. Thank you, Paul. Thank you, Steve. And on that note, let's wrap the symposium up. So thank you both again, Steve and Paul for a very insightful discussion today. I learned a lot. And we have more to come. So the spring lineup is shown on the screen here. We have a lot of great speakers, very diverse from different part of the innovation pipeline, as well as sector. Next, in two weeks, we have a session on new aqueous chemistry from Chunxing Wang, also from the University of Maryland, each in lieu from the Chinese University of Hong Kong. Two weeks after, we're going to have Tim Holm, the CTO and co-founder of QuantumScape and Jennifer Root from MIT speak about solid state batteries. And after that, we're going to have a session with Diane Grinich who is formerly the commissioner of CUPC to talk about building energy efficiency and energy storage. Then we're going to have Frank Blum who is the head of battery at Volkswagen, which has just announced a major initiative in EVs. And then to wrap up the quarter, we're going to hear from Yankuk Sun and Hubert Gastagers on the latest chemistry developments for lithium-ion battery cathodes. So I hope you will mark your calendars and join us in these coming sessions. Thank you everyone and have a great day.