 My name is Will Chu. I'm the Faculty Code Director of Storage X Initiative and a Professor of Material Science and Engineering. I'd like to welcome everyone to today's Storage X symposium. Over the past 18 months, you have heard a lot of talks from speakers working on technologies, working on chemistries, working on new approaches for energy storage, and we have also discussed a number of systems level considerations for energy storage. And today we're going to have a very special talk that addresses energy storage at a whole systems level to enable decision making, considering many factors. And importantly, the talk will focus on technology, agnostic approach, therefore allowing the analysis to be applied to any energy storage, whether it is electrochemical, thermal, mechanical, and so forth. And to give the lecture today, I'm extremely pleased to introduce a rising star in the area of energy analysis, E.J. Beck, who is a graduate researcher in the Department of Energy Resource and Engineering. And E.J. is currently completing her dissertation with Professor Sally Benson. I'm so pleased to welcome you, E.J., to speak to us today. The floor is yours. Thank you so much for that kind introduction, Will, and let me share my screen and get started. Hello, everyone. Thank you so much for having me here today. Today I'm excited to discuss synergy and competition between long and short duration storage and renewable power grids. And I also want to make sure to acknowledge my colleagues in this project, Andrew and Sally as well. Now, before we start and go into too much depth, I wanted to emphasize the key takeaways from this analysis. We utilized a detailed electricity system model to assess short and long duration storage operation and value in renewable power grids. We find that in a high renewable grid, energy storage frequently serves 40 to 50 percent of grid demand, really emphasizing the importance of energy storage in future decarbonized systems. Now, we find that the utilization of short and long duration storage actually have a positive correlation that is quantifiable by the average state of charge. And I'll go into a bit more depth in that. Now, when short and long duration storage are operated synergistically, system costs can be cut as much as half relative to a scenario without any long duration storage resource that exists. Now, we believe that approximately a $5 per kilowatt hour is an appropriate benchmark for long duration storage to operate synergistically with short duration storage and really start contributing to deep decarbonization in a cost effective manner. And finally, a system with long duration storage and clean from generation resources results and the lower system cost. So I'm sure that the audience today recognizes the importance of decarbonizing the electricity sector. Now, deep decarbonization of the electricity sector will require significant capacities of clean variable renewable resources, such as wind and solar. Now, as you can see on the figure on the right, energy storage is used to time shift variable renewable energy resources to times of higher load. And energy storage is the technology that'll help incorporate higher shares of VRE within the grid. And because of this, energy storage is a critical technology for deep decarbonization. Now, I do want to note that the timeshift needed ranges from anywhere from minutes to hourly basis all the way to seasonal timeshifts. And this is based on the seasonal intermencies of these variable renewable resources as well. So this is an illustrative figure that really just shows how the maximum required storage duration of the Y-axis increases with increasing shares of electricity from wind and solar. And it really highlights that as a share of variable renewable resources and the grid grows, so does the need for longer duration storage. Notably, grids with more than 80% of winded solar shares will likely need storage on the order of weeks and seasons. What type of storage resources are out there? And based for today's analysis, I'm going to generally characterize them into two different categories, short and long. Now, the duration of storage is basically the ratio of energy capacity to the power capacity of the research, which in other words is the time it takes for the resource to complete a full discharge at its maximum power rate. And this table briefly compares short and long duration storage. Now, short duration storage is generally less than 10 hours in duration and at most, sorry, at most 10 hours in duration. And as a result, focuses on time shifting generation diurnally. Short duration storage is generally more mature and already deployed today, most typically in the form of lithium ion and lead acid batteries. Short duration storage has relatively moderate energy and power component capital costs as well. Now, on the other hand, long duration storage ranges from 10 to thousands of hours in duration. And so long duration storage is able to smooth weekly and even seasonal fluctuations in generation. Now, while long duration storage technologies are largely nascent, some potential technologies include hydrogen gas storage and compressed air energy storage. And examples of existing longer duration storage include pumped hydropower. Now, finally, long duration storage typically has much lower energy capital costs, which is why it's able to have a much longer duration, but they also have significantly higher power capital costs. And so modeling these two storage resources separately is really important in understanding their distinct roles and net zero carbon grades as their different price components and their durations greatly impact the role that they play. And I just want to note that throughout this presentation, a lot of the acronyms that you'll see on the screen, short duration storage will be represented as SDS while long duration storage will be represented as LDF. Now, a lot of existing literature today, as well as really most of the developed capacity of energy storage in the US in recent years have largely focused on shorter duration batteries. So the figure shows the power capacity and duration of large scale battery storage by region of the US. And you can see that across all regions, most of the duration of existing energy storage resources typically fall at less than four hours. But as we continue to decarbonize the grid, there is a growing recognition of the importance of longer duration storage resources. And so recently, the US Department of Energy, the DOE, just announced the long duration storage shot. And the long duration storage thought basically established the goal to reduce storage costs by 90% in storage systems that'll deliver 10 plus hours of duration within the next decade. The long duration storage thought is really, it considers all types of technologies, it's technology agnostic, whether that's electrochemical, mechanical, thermal, chemical, or really any combination of those that has the potential to meet the necessary duration and cost targets for grid flexibility. Now, if we take a look at the existing literature on system wide assessments of long duration storage resources, there've largely been two types of studies. The first type represented on the left here utilize isolated type of economic analyses to establish benchmarks for development. So the example on the left taken from Schmid et al. shows what types of energy storage technologies have the lowest cost for different use cases of discharges per year and hours per discharge. Now, while these analyses are incredibly helpful in comparing longer different long duration storage technologies with short duration storage, it doesn't provide much insight into how these roles would fit within the context of the system operation. So the second type of studies on the right side here utilize macroenergy system models to assess the system value of long duration storage. And this is an example taken from Dowling et al. that shows how system costs vary for different energy storage costs. Now, together, these studies have been incredibly valuable in assessing various energy storage technologies. They've also highlighted the seasonal and even multi-ear storage roles that long duration storage could play, making it an incredibly valuable resource and least cost net zero carbon electricity system. Now, I want to note that these studies have highlighted that at energy storage costs of approximately $20 per kilowatt long duration storage would have value in decarbonizing the electricity system costs effectively. And I'd like the audience members to kind of keep that number in mind as we go through our own analysis. Now, what's missing from these analyses, though, is the insight into the operation of long duration storage resources and their interaction with other storage resources. And the reason we need a better understanding of this is multiple. First, technology developers and researchers need a better understanding of the use case for long duration storage that will help inform future research and development programs. And specifically for technology developers, they really need to understand how long duration storage would operate in the grid because they have to design and optimize their technologies to meet those specific needs, whether that's based on specific discharge patterns or energy capacities or power capacities or durations. Now, policymakers need to recognize the value of long duration storage and its role in decarbonization to ensure that long duration storage is appropriately reflected and included in ongoing climate policy. Now, policymakers will also have to establish appropriate market functions that would enable a profitable economic operation of energy storage with the specific operation patterns that we might see throughout this analysis. And finally, policymakers will have to anticipate future grid dynamics for infrastructure development, which is becoming ever more important nowadays, which really might vary depending on the presence and utilization of long duration storage resources. And so altogether, our analysis fills this missing gap and addresses these points by focusing on understanding the operational behavior and associated system value of long duration storage in conjunction with short duration storage using a very, very detailed capacity expansion and dispatch model. Great. So now I'll briefly go over our method and approach. Now, we focus on California as a case study because, well, one thing, Stanford is here, but also because it is one of the states that has a 100% clean energy grid goal by 2045. So the modeling implications of our analysis actually bears some blend to a possible pathway for decarbonization for the California state. And also, California has an abundance of renewable generation resources within the state and in particularly PV. So it's a very, very good case study to explore options for high renewable grid. The main tool that we use is called a capacity expansion and dispatch model. And it's fundamentally an electricity system model that co-optimizes the cost of building new generating and storage resources, as well as the cost of operating the entire system. And I'll briefly walk everyone through how this model works. So we start with existing power plant capacities in California. We're not starting from scratch. We're starting from kind of the existing capacity that we have today. And we also have to input very detailed hourly electricity demand profiles for the future in 2045. Then we also have to collect several different solar and wind generation profiles across the state. So we compile all of this information and data and we input it into our model. And we also have to consider a wide range of techno economic parameters, including the capital costs or the investment costs of building new generation and storage resources, as well as their respective operating costs. We of course have the ability to model different policies, so whether that's a renewable portfolio standard or an emissions limit. And in this case study, we are modeling a net zero carbon grid. So then based on this, the model minimizes the total system cost, which includes capital costs, ONM costs, fuel costs. And in addition to minimizing the total system cost, it simulates the grid to satisfy the demand that is given on an hourly basis for the entire year. And the largest constraint of the model is that really for the newly built system, the demand has to be satisfied every hour of the year, ensuring that the system built is a reliable one. And so for the types of outputs that we'll get and the types of results that you will see throughout the rest of this presentation is of course the total resulting system cost, the total system capacity for both generation and storage energy resources, as well as the hourly operation of each of these resources as well. So just to bring this all together, what we're doing is building a future California grid in 2045 that is net zero emissions. And each model run that we do represents a minimum cost system given the assumptions that we are putting into the model. The system that is built satisfies the demand for all 8,760 hours of the year that is modeled. Now, for simplicity's sake, within this analysis, we only allow the future expansion of onshore wind, offshore wind, solar PV, and storage. And I do want to note that onshore wind and offshore wind capacities in California are relatively limited right now. So the bulk of the new generation build that you'll see comes from solar PV. For energy storage, we modeled two representative energy storage technology, one long and one short. So short duration storage is modeled as a four hour storage resource with 85% round trip efficiency with capital cost of $100 per kilowatt hour, which is quite consistent with optimistic declines and future costs for lithium ion batteries. For long duration storage resources, we actually model of quite a wide range of both energy and power investment costs or capital cost components. And the range is indicated there on the bottom on the left. Now, as a result of varying the energy and the power cost components, the duration of the long duration storage is actually optimized within the model. And the only constraint that we put it is that the duration has to be longer than 24 hours. We model long duration storage with a 45% round trip efficiency. So these are the two representative technologies that we'll be modeling. And with these resources, we model six different case studies. And these case studies are intended to assess the dynamics of the system across a wide range of varying assumptions on the grid composition itself. And so first we have, of course, our base case scenario. Now, the second is a wind scenario in which we allow offshore wind to be available at a lower price. And so there's more offshore wind that's built within the system. The low cost short duration storage scenarios consider short duration storage at $50 per kilowatt hour instead of 100. The high round trip efficiency scenarios considers long duration storage resources at 85% round trip efficiency relative to 45. And the 100 hour scenario limits the duration of long duration storage to be less than 100 hours. Now, finally, in the carbon capture and storage scenario, we allow an additional generation resource, which is gas power plants equipped with carbon capture and storage, to be chosen as a resource as well in the mix. Now, CCS is what we call a clean firm generation resource, which is a clean resource that is dispatchable at any time of the year without any constraints from the weather system. And so this case in particular is intended to assess how long duration storage resources interact with are similar to or differ from a clean firm generation resource. Finally, let's get into our results. So first, this figure is a box on whisker plot of all the scenarios that we ran across the different case studies. And what it shows is the percent of annual load that is met with energy storage both short and long. And the box on whisker plots are the results of the 32 different long duration storage price point scenarios that we ran for each case across the access. And so what you see this figure fundamentally shows is that in a high renewable grid, energy storage frequently serves 40 to 50% of the grid demand, the range depending on the price of long duration storage. Now, this is a quite a significant share because if you think about it, it's basically equivalent to the share of natural gas generation in California today. Now, the cases in which energy storage does not serve 40 to 50% of the load is when other sources of generation such as an additional offshore wind and CCS are available within the grid. Now, let's divide this percent of load served by short and long duration storage and take a look at that. So on the left here, we have the same data, but specifically for long duration storage while on the right side, we have it for short duration storage. Interestingly, across most cases and price points, long duration storage actually serves less than 10% of the load on an annual basis. And this is relative to short duration storage that meets 30 to 45% of the load, and it's really largely responsible for the majority of the energy coming from storage. Now, the only case in which long duration storage plays a larger role is the high roundship efficiency scenario. And we'll get into a little more detail later on why that's in. So now that we've seen how much of the load is met with each resource, let's take a look at how these resources or energy storage resources are actually meeting these loads. This plot here shows a weekly generation profile for a representative winter week for the base case scenario under three different long duration storage energy prices of .5, 15, and $30 per kilowatt hour on the bottom there. And the most notable resources that might pop out are the long duration storage resources in blue, the short duration in yellow, and the PV generation in dark orange. And so those little kind of orange peaks are when the sun is up during the day. Now, when long duration storage is affordable up on the top there, we see that it's almost playing a base load resource, particularly for days when PV generation is low, while short duration storage plays an hourly to a diurnal role, where it operates by charging during the day and discharging when the sun is setting or during the nighttime. And I do want to note that even though we call it a short duration storage resources and limited to four hours, the low cost of $100 per kilowatt hour really enable that short duration storage to be built in bulk and play a diurnal role in meeting most of the nighttime load instead of just shifting load on a per hourly basis. And so it's quite playing quite an important role in this future grid. Now, as long duration storage gets more expensive, we see less and less of long duration storage, but more and more of that yellow short duration storage. And notably, as long duration storage gets more expensive, the PV capacity grows significantly as well. And what this is showing is that when affordable longer duration seasonal storage is not available, a combination of significant capacities of short duration storage and PV is needed to ensure reliable operation throughout the year. And most particularly, at least in California, it's really needed in the winter time when PV generation becomes much lower. Now, let's expand this generation profile to all other six case studies that we ran. And it's a little more crowded, but basically each row is showing the different cases we ran, while left to right, it's increasing long duration storage prices. So we see quite similar behavior across most scenarios where as long duration storage gets more expensive, PV capacity grows and short duration storage is really being utilized diurnally every day. Long duration storage, on the other hand, is used quite sparingly. Now, another interesting point to note is that when there are other generation resources such as more offshore wind and carbon capture and storage available, there's much less overbuild of PV capacity on the right-hand column there. And what that indicates is that with a wide range of generation resources, this can prevent the overbuild of a single resource and help build a more flexible grid in adjusting to feature discrepancies and leathers. Now, this figure and this analysis so far has showed the operation of long and short duration storage resources on an hourly and a daily basis. And let's take this, take a look at how they operate on a seasonal basis. So these 3D plots here show the annual state of charge for long duration storage and short duration storage for three different long duration price points, which is actually consistent with the price points that we saw before. And the way to read these storage plots are the bottom right axis is the day of the year, while the left axis is the hour of the day, and the vertical axis shows the normalized state of charge for the storage resource. So let's take a look at that figure on the far left as an example. For long duration storage, there isn't much variation in charge or discharge on a daily basis, but there is a huge variation on a seasonal basis. So you can see that it's varying a lot throughout the season. So long duration storage is being used seasonally, charging in the spring and summertime, holding that charge throughout the fall, and depleting in the winter. And this is very consistent with California's weather, as I mentioned, when PV generation is very low in the winter, and that's when long duration storage is needed and is being utilized. On the other hand, for short duration storage, we see a huge variation on an hourly basis. You see that short duration storage charges during the daytime and it depletes it overnight. But on a seasonal basis, it really doesn't vary at all. It's consistently charging and discharging to its full capacity each day. But as longer duration storage gets more expensive from left to right, you see that long duration storage is used more and more sparingly, really only for a couple of weeks and days in the winter time. And for short duration storage, while we still see the diurnal pattern, short duration storage has to start playing a more seasonal role. And so we actually start seeing a lot more of the seasonal variation for short duration storage. Now as a result of this variation in operation, we find that the average state of charge that the entire year of both of these resources is actually a really great indicator of these different behaviors. And so for each of the 3D profiles, we've listed the average state of charge on the bottom there. And so as long duration storage is used more sparingly, you can see that the state of charge goes from 0.56 to 0.06. While as short duration storage is used more seasonally, the short duration storage state of charge goes from 0.5 to 0.26. And so fundamentally what it shows is a lower state of charge, average state of charge indicates a kind of less utilization of the full capacity of the resource. Now if we take these results, so for each scenario there's a average state of charge for the long duration storage and an average state of charge for the short duration storage. And if we actually plot this, this is what it looks like. So on the x-axis, we have the average state of charge for short duration storage and the x-axis, we have it for long duration storage. And notably, we actually see that there's a very strong positive correlation. And what that means is that when one resource is utilized efficiently and utilized well with a high average state of charge, the other one is utilized very efficiently as well. And this is because in those cases, both energy storage resources are very well operating within their suitable role. So long duration storage is playing a very seasonal role while short duration storage is operating direnally. But as we move to the bottom left there, when one storage resource becomes underutilized, and in this case, when long duration storage basically becomes used a little more sparingly, we see that short duration storage is pulled into this imperfect substitute role where it really has to overbuild itself and start playing a seasonal role. And so the average state of charge for short duration storage decreases as well. So it's really remarkable to see this positive correlation in the utilization of both short and long duration storage resources. So let's take a look at that same figure. But now what I've done is divided this into four different quadrants. The four distinct quadrants really show distinct operating regimes. And the division between the regimes are simply the averages of the average state of charge. So it's it's just a really a nominal kind of descriptive division that we put there. But the four descriptive operational regimes are on the top right, we have the one in blue is synergy. And that's when both resources are utilized efficiently. The green, which is LDS dominant or long duration storage dominant is the one where long duration storage actually dominates all energy storage function and displaces a large share of short duration storage. So that's when you actually see a lot more of the competition between the two resources. This is something that we actually never see happening in the base case scenario. And I'll explain briefly later in a while. Now the top left is a long duration storage underutilized scenario in which short duration storage is utilized pretty efficiently, but long duration storage starts being underutilized. And finally, the bottom left scenario in red there is when both storage resources are underutilized. And what we find is that the operation of the system within each of these operational regimes is pretty consistent. Another interesting fact is that now in addition to the base case, if we take a look at all of the rest of the cases we ran, there actually is a very strong positive relationship that persists across all the scenarios. Now it is a little weaker for the high round trip efficiency and the carbon capture and storage scenarios. Particularly for the high round trip efficiency scenario, the distinct roles of short and long duration storage become a lot less clear because long duration storage becomes much more competitive with short duration storage based on its round trip efficiency. So that's the scenario in which we actually see that a dominant long duration storage resource leads to overall higher utilization, shifting a lot of the scenarios into the synergy regime. But there are also of course some scenarios in which the long duration storage is actually dominant and so there's some scenarios there which where most of the short duration storage is actually displaced. Now for the carbon capture and storage scenario all the way on the right, the presence of a clean firm resource fundamentally reshapes the role of storage overall. We see that short duration storage is better utilized in playing more of a diurnal role to kind of compliment the generation from clean firm resource and it generally decreases the role of long duration storage because there is this resource that's readily available for all seasons. So as a result of that we see an overall higher average state of charge for short duration storage. Right so this is so far we've looked at how these different operating regimes map up onto the scenarios that we ran and now what I want to do is show how these operating regimes actually impact overall system costs. And to do this we ran a reference scenario which is basically a base case scenario without any long duration storage available. So that'll be our basis of comparison for costs. So if we overlay the resulting system costs of all the scenarios and cases revan and here we're showing it as a percentage of the reference cost scenario and color coded by the operating regimes you'll see that the operating regimes are highly correlated with system costs. So ultimately synergistic scenarios which are blue are much lower in system costs relative to the red the all storage underutilized cases. And you'll see that the long duration underutilized scenarios tend to be more expensive than the synergistic scenarios as well that's the one in orange where it kind of lies in between. Now I do want to point out that having long duration storage at any price point really does contribute to reducing the system costs of decarbonization and that's very consistent with a lot of the literature that we've seen. But what this shows is that when both storage resources are operated synergistically that's when system savings are the greatest and the system costs are cut in half relative to having no long duration storage at all. And so overall we find that these operating regimes are quite accurate proxies for system costs. Now this is a figure that now takes a look at these operating regimes based on the long duration storage costs that were modeled. So the x-axis is the different long duration storage energy capital cost that we modeled and the y-axis is the very power cost that we model and each little box there shows the resulting operating regime. And you'll see that the operating regimes are largely driven by the cost of long duration storage and particularly the energy cost of long duration storage which means that you can see that the x-axis value matters a lot more than the y-axis. And this means quite intuitively you know that the more affordable long duration storages the more synergistic operation that we would see. And on the other side more expensive long duration storage results in underutilization of all the resources. Now I've added this little red dot there as you'll see on the five dollar per kilowatt hour benchmark and we believe that this is kind of the benchmark in which long duration storage results starts operating within that synergistic regime. And I want to mention I noted earlier in the presentation that previous literature assistant found that approximately twenty dollars per kilowatt hour could serve as a benchmark for long duration storage to contribute to cost effective decarbonization of the grid. And our analysis shows that yes while that is reaching five dollars per kilowatt hour can really optimize the system so that all the resources are utilized efficiently and the system costs to reduce the most. And as a reference just to kind of give a context of what that five dollars per kilowatt hour might mean. Current literature on long duration storage technology shows that currently viable technologies for that price range might be geologic storage of hydrogen, compressed air energy storage and potentially some forms of thermal storage as well. Now if we take a look at the rest of the scenarios that we ran you can see that the benchmark for five dollars per kilowatt hour is more or less really persistent across all of those scenarios. But if longer duration storage is limited to less than a hundred hours in duration you'll see that the prices for long duration storage have to be much lower for that resource to operate within the synergy regime. And so it's really important that we do have you know a pretty long duration in addition to the lower cost to ensure this operation in the synergistic regime. Now this is the same figure I've just moved it up there a little bit. Now let's take a look at this figure in relation to other variables that are our outputs within this model. So this figure on the bottom has the same axes as the figure above so we're showing system cost as a percentage of the reference scenario. And so these are the same results that we just saw before in a cost curve but we're showing it in an interpolated kind of color color color here. And very clearly you can see how the operating regime matched up very nicely with system cost which was shown before in the cost curve. Now in addition to this system cost though let's consider another variable or result and that is system generation capacity. And what we see is that the operating regime correspond very well with the overall generation capacity that's built as well meaning that the synergistic operation of energy storage resources actually results in the least capacity build as well as you can see on the axis center. And so for the scenarios where both energy storage resources are underutilized the generation capacity can be more than three fold larger than the the model peak demand that we have. And this really raises an interesting question on the relationship between system cost, generation capacity, and storage capacity. And the question is what's really driving the system cost to be higher? And so this next figure kind of gets to that point and what it shows is solar PV and short duration investment cost. Sorry this figure really shows that solar PV and short duration investment costs are really the drivers of system cost. So the three figures we have here show the investment cost for long duration storage, PV, and short duration storage. And the system cost is shown on the Y axis there and the scenarios are color coded by the operating regime. And just as a gentle reminder as I put there synergistic scenarios are driven by lower cost longer duration storage resource resources. Now what's interesting is that the investment cost for long duration storage actually doesn't really show a clear trend or relation to system cost. On the other hand we see that both solar PV investment and short duration investment has a very strong positive correlation with system cost. And if we try to understand why this might be the case this is a figure now I've got it on the left that shows the amount of long duration storage energy capacity that's built in the system. And here we actually do see a pretty linear trend meaning that the more within the synergistic regimes there's a lot more capacity built and in the underutilized scenarios there's a lot less of the resource built. And so in comparing this with the long duration storage investment scenario and synergistic regimes that which are the blue kind of dots there the relationship with system cost is positively linear. So as more capacity gets built the investment cost increases as well. So that's moving up to the right in the cost there up here. But once long duration storage starts getting pricier and they start being underutilized less and less capacity of it is built. And because of that the overall investment cost actually decreased to the left there. And so as we saw before as the long duration storage capacity decreases more short duration storage and PV capacity is overbuilt to make up for the absence of long duration storage. And this is fundamentally what drives the system cost. It's you know driven by long duration storage but ultimately the direct cost drivers are PV and short duration storage investment. Now one last point before we wrap up. I want to come back to this figure here because I don't know if someone noticed but I didn't include any of the CCS scenarios here. And so now I'm going to add in the CCS scenarios to the right. And so you'll see on the top right figure there that the long duration storage in the CCS scenarios behave consistently with the other scenarios meaning that when it's cheap it operates synergistically and when it's pricey both storage resources are underutilized. So that trend there persists. But what's really notable are the system costs and the capacities there on the bottom. I want you to note the different scales for the CCS scenarios. So even the most expensive CCS scenario is about the same price as the most cost effective scenarios without any carbon capture and storage. Now there's been a lot of research already that shows the value of clean firm generation resources. You know these are resources that are clean and dispatchable at any time of the year as needed. So despite the often higher capital costs and actual fuel costs for operating these resources they're incredibly valuable for ensuring reliability throughout the entire year which is why the system costs as well as the generation capacity build are so much lower. And so long duration storage energy costs will definitely have to decrease significantly to directly compete with such resources. But what this shows is that including both long duration storage with the clean firm generation resource is fundamentally what results in the most system cost savings. And it really highlights the value of having the most diverse set of generation and storage resources. Now with that I'd like to briefly summarize our talk today. I showed an analysis that utilized the detailed electricity system model to assess short and long duration storage operations and their value in renewable power grids. We showed that in a high renewable grid energy storage takes a bulk of the grid demand or meets bulk of the grid demand really emphasizing the importance of it. And we find that the utilization of short and long duration storage has a positive correlation that is quantifiable by the average state of charge. And when short and long duration storage resources are operated synergistically system costs can be cut in half relative to a scenario without any long duration storage resources. We believe that based on this analysis a five dollar per kilowatt hour benchmark is really what a long duration storage should strive for to operate synergistically with short duration storage. And finally a system with long duration storage and clean firm generation resources results in the lowest system cost. And with that I will conclude my presentation and I look forward to any of the questions. Thank you very much. EJ, thank you for the fantastic talk. And let's go ahead and jump into the Q&A. So maybe I will start with a first question at a higher level. So at the end of the talk you discuss alternatives to energy storage such as Peaker with CCS. And I wonder if you can broadly comment on maybe the two scenarios that you didn't present, which is what if you, I know this is not possible in California but as a consideration, what if you build out more variable generation such as wind that has more seasonal characteristic? I think that's one scenario I'm very excited to hear about. And the other one's nuclear as a way to contribute to a lower footprint baseload. Can you comment on how those two might behave just from intuition? Yeah, great question. So first on wind we tried to get to that point a little bit with the offshore wind scenario but obviously there are grids out there that are much more wind dominant than the PV that we've seen today. Literature has shown that the relationship with short-duration storage and PV is much stronger and the correlation with long-duration storage and wind is much stronger. And so I feel like in the heavier wind scenarios we'll see probably less short-duration storage built, more of it being utilized less on a diurnal basis but more really to fit those short-time shift needs from based on wind. But also a general decrease in longer duration storage resource needs as well just because there is more kind of seasonal operation of wind. And so in a way you know as we saw consistent with our results having offshore wind or any large wind generation might generally decrease the role that storage might play but also they might be very important in different ways. Now nuclear is actually an important one too. So in my previous work other than storage I've actually taken a look at different types of clean firm resources so nuclear being one of them. And so just to point out the difference between nuclear and CCS the capital cost is much higher the variable cost is much lower so it operates more as a base load resource. But what we actually find from previous analyses is that it doesn't really matter what type of clean firm resource is utilized in the grid they're all pretty cost effective in decreasing the load. And the way nuclear does that is by you know displacing a lot of the variable renewable generation needs within the grid because it's so cheap for its energy. Now the cost savings with CCS and nuclear are pretty comparable. And so in a way I actually don't see having nuclear change the results so much from the CCS scenario that we saw. They're both clean from generation resources they'll help California get through those winter months very effectively and reduce the overall burden that these variable renewable resources and storage resources have to play. Thank you G. Just to make sure I understood your points here. So essentially when one considers long duration storage the competition so to speak also comes from other base load and variable generation options that could have a very low or negligible carbon footprint. I think your CCS scenario really shows that very very well. Yeah I think that is an accurate way but also to emphasize while they do compete having all of them really does result in the lowest system cost as well. So again going back to this whole synergy and competition theme there definitely is a little bit of both going on there. Terrific. E would you like to ask the next question? Yeah sure hi EJ very nice talk good analysis right here. I want to ask you a couple of questions to start right so from my side. One is on the energy efficiency. I think your assumption is of course very recent of 85% lithium iron 45% for long duration. This is looking at some existing technologies. I'm wondering for long duration storage what if the energy efficiency can go higher let's say 80% might even be 85%. Would that change the whole scenario quite a bit? Yeah that's the first question maybe I'd like to answer the first one. Yeah yeah that's a great question. So we did look at that scenario with the higher round trip efficiency at 85% and from my from based on what I've seen that really is the scenario that changed the operation and the dynamic of the storage resources the most. And so you're totally right and that having a higher round trip efficiency for long duration storage makes it a lot more directly competitive with short duration storage and so you definitely see in some cases when long duration storage is cheap enough it actually becomes a more dominating storage resource within the grid. Yeah so well if I may I ask my second question so you have this framework right here really nice so we can explore the parameter space a little bit more and I can see in the audience there will be a lot of people say hey based on this guy like we know where's the button to push very important things to improve in the technology. One thing is also related to you know long duration storage like the kilowatt hour could be low cost you have different number right there but the power cost is high let's say kilowatt if it's a thousand dollars of course the cost is high. I'm thinking about a scenario likely to be coming in the coming decade is you know if you have in the battery language is called C-rate if you can charge this charge at C over 10 right particular discharge in this case more relevant might even be C over five so if you could get to kilowatt hour cost that's five dollars or ten dollars let's use ten dollars right I can do C over five that means my power cost I could handle easily you times you know five times or ten times five that's a fifty dollars per kilowatt that's very low cost of power would that change the whole equation quite a bit the whole scenario just say wow it's not a thousand dollars right you know maybe I make it a little bit more expensive I can get to a hundred dollars per kilowatt or two hundred dollars per kilowatt that probably you can change it thanks so much yeah that's a that's a great question and we actually did run additional scenarios that looked at this point we didn't go quite as low as 50 but in comparing we recognize that there is some trade-off and I don't know if this is quite what you're getting out but I did want to mention that there's some trade-off in getting higher efficiency resources and the power cost because if you invest in the power cost a little more you might be able to increase the efficiency but from increasing power costs and so we find that increasing efficiency is actually slightly more effective at lowering overall system cost than the overall power cost and obviously in this analysis we also saw that energy costs matter a lot more than the power cost now that is all to said that's all within the context of the lowest power cost that we modeled being 500 dollars per kilowatt hour now for 50 dollars per sorry 500 dollars per kilowatt what's interesting is that in the scenarios that we saw most of the energy composition of energy storage so that's the total like per watt hour gigawatt hours of storage built the bulk of that comes from long duration storage and the bulk of the power dollar per kilowatt or the kilowatt gigawatt basis comes from short duration storage and so yes I can definitely imagine a scenario in which when the cost of long power cost of long duration storage decreases the long duration storage actually fills up more of that overall power capacity needs of the grid as well and starts displacing short duration storage so point being yes I can see that dynamic starting to displace short duration storage I'd be very curious to see by how much it displaces it especially depending on the efficiency of the system but overall if there are three variables within long duration storage which is power cost energy cost and efficiency at least in the variables that we modeled power cost actually matters the least overall in reducing system cost yeah I guess what I mean is right there eventually is actually your long duration cost can also do your short duration cost right well it's kind of coupled so strongly together so I'll pass back to Will thank you Yashua was going to ask a very much related question um so Yashua just to repeat the point he made in a different way so it's hard to use short duration storage technology for long duration storage but the reverse is not true as Yashua's pointed out and I'm curious to understand the tipping point right so I think if you were to plot many of the proposed long duration storage technology have relatively low round trip efficiencies and so forth so this is a trade-off in order to get the low dollars per kilowatt hour however as you pointed out the long duration storage requirement are seasonal so you know the time of charge right and the time of discharge might be quite fixed on the calendar right so you will not be using it during the summer so have you consider or maybe you can comment intuitively in those months in which you're not using long duration storage for long duration storage can I basically view it as just free to operate for short duration storage um in those months and will that change the equation at all because your capacity will be huge in terms of kilowatt hour installed for long duration storage so I put it another way if you have a long duration storage of say let's just say one week right so I can certainly use it for a day in a very shallow cycling my guess is that it doesn't add very much to the cost but can you help us understand the benefits yeah that's a that's a great question and maybe I'll provide a little bit more context on the specific model runs and why the results that we're seeing are the way we are because basically what's happening is that I think the short duration storage costs that we modeled are quite optimistic uh there's a lot of um high efficiency and so generally the model shows that that is the preferred kind of overall bulk use of storage and so definitely once once it starts um so okay so that's the preferred use of storage and so because it's so cheap what the system is building is that you know even though it might be the not the most efficient kind of duration build out it's just building a lot of it because it is cheap to overbuild it and so that's why we see this long duration storage role become smaller and smaller uh not necessarily because of um you know it's it's not cheap to operate but more so I think it's driven by the low cost of short duration storage so to your question if we do see a scenario in which maybe the the short duration storage is built a little more moderately we're not necessarily talking about a net zero carbon system the system has a little more flexibility than 100 yes I think long duration storage can definitely be utilized on that short duration basis as well um I think that's just a function of the model the way it's set up it's it is you know we're only considering a year in large hourly time steps and so you don't see as much of that uh interaction happening but I could see in reality that that being the case as well terrific is yeah I think it'd be great to understand the tipping point in terms of the round trip efficiency right so if the round trip efficiency is incredibly low for long duration then it's not going to cut it for short duration uh because the energy cost would be too high but I think as E was pointing out if the the efficiency is reasonable then there should be a optimal composition of the short versus long well that basically the utilization of long duration storage for long duration storage I think that could be a quite exciting quantitative analysis to perform yeah and and I do want to highlight that the the 85 round trip efficiency scenario that we ran if we actually look at I don't know if I can share slides again so if we actually look at the high round trip efficiency scenarios when it's low cost uh we do see it operating really mostly as the main storage resource where there's barely any uh short duration storage resource so we can imagine actually the tipping point if long duration storage is 85 percent round trip efficiency the tipping point being somewhere in between point five and fifteen dollars per kilowatt hour because we do see that at least point five dollars per kilowatt hour it is really the dominant energy source for both short and long duration operation of storage very exciting um let me ask a slightly different question and and E thank you so much for incorporating all of the viewer's question into your question so I cover quite a quite a number of them so you chose a few scenarios I think with fairly limited set of technology attributes I was wondering if you have performed a more continuous sensitivity analysis on the technology attributes themselves and the question I'm trying to get at is what would be your recommendation to technology developer into what they should be aiming for so we're just talking about round trip efficiency could be an important knob and to really sort of understand you know how good is good enough right or the optimal point um of say cost and performance metric balance so you talked about round trip efficiency and others are are there other parameters you think technology developers should be looking at in terms of making it more attractive other than just the cost aspect yeah that's a great question maybe one thing is that the duration and I know I think duration is really interesting because I didn't mention this too much in depth but a lot of the scenarios that we do see in the analysis have very very long duration storage resources 100 hours plus and now the definition of long duration storage from myself during the literature research the range is very wide from 10 hours to 1000 hours so you know a 14 hour resource might be considered medium long duration but even a thousand hour resource might be long duration resource and so I think the one scenario where we ran the case where the duration was limited to 100 hours actually showed quite different dynamics where that long duration storage isn't able to sufficiently play a very reliable seasonal role it still relies on overbuilding generation a little bit to ensure reliability throughout the entire year and so you know cost is important but unless we are actually able to develop a certain storage technology that is able to you know hold that energy storage for longer durations at a time I think um as we showed the cost you know doesn't matter a little less for those scenarios because the duration matters so that might be one other parameter that I might think about for developers great thank you DJ E yeah I um you know now if you look at a scenario right so we're getting more complex a little bit more complex I'm for quite a while I mean this is still holding true this combination of the storage if we could combine carbon capture sequestration the cost is low enough it would be interesting EJ to see this right if you get to $100 per ton $50 per ton and $20 per ton of CO2 capture you know cost that that whole thing then the natural gas serve as reasonable base low normally base low maybe serve as a seasonal storage right how this coming in will be really interesting because you work with Sally because you guys must be thinking about this every day right so big way to see your inside a little bit you know giving the CO2 captures cost yeah yeah I think what we found is that carbon capture and storage or any in this case gas of carbon capture and storage is really it's a generation resource which is what makes it very different from these long duration storage resources and so in the previous analyses we've had even at very high cost of carbon capture and storage and we've also modeled cases with nuclear with a capital costs are much higher and we've also actually modeled cases with a form of zero carbon fuel where the variable costs are much higher and so these are all different types of clean firm resources with varying cost components or more expensive capital costs more expensive variable costs and what it fundamentally shows is that they operate differently but they operate to provide that reliability needs throughout the winter time and don't necessarily depend on other generation resources to do so and so you know while the cost sensitivity might be interesting having done that a little bit it really doesn't alter the value of these resources or the system costs as significantly I think long duration storage it's almost twofold in why the cost drivers impact the system costs so greatly it's because you know in any amount of long duration storage we build or can't build because of the price it's overbuilding a lot of other resources to make up for that generation as well and so I think that's why we see a lot more of the sensitivity system cost sensitivity to long duration storage prices relative to what I've seen from carbon capture and storage prices yeah thank you will now this is fantastic Ej learning so much here let me also try to incorporate some of the questions from the audience into my next question so as I understood it Ej in your analysis you're accounting for the reliability of the grid but are you also accounting for these exceptional events and the costs associated with them and certainly the options you have talked about all address that differently right can you come in a bit more about the resilience issues to extreme events yeah that's a great question and I'll actually pull in some of the insights that I've gotten from previous work that touches on that so these models are only really modeling the system for a representative year and so no we aren't modeling any extreme weather events or you know strenuous circumstances that go beyond the typical year in a previous analysis what we did was we had a model run a single weather year versus multiple weather years that kind of encompassed a different you know bad draw ears where the hydro generation was lower the or some weather the pv generation was very low and compared kind of the resulting capacity that comes out from having one year and multiple weather years and the scenario with the multiple weather years required on the orders of hundreds of giggle like 200 gigawatts more of pv capacity to ensure reliability when just relying on energy storage and these intermittent resources now I will say this is actually case where we didn't model long duration storage so it might be a little I definitely think the reliability of the system would be much greater with long duration storage but there definitely is an impact in considering these more extreme events and so if anything I think this would this would be a more conservative estimate of what we imagine the grid to be if we want to go to net zero carbon now I will say I'm sorry with with the scenarios that we ran with queen firm generation resources with and without with multiple weather years and with a single weather year that showed a lot more stability in the overall capacity build again because these dispatchable resources were available year long and so I can imagine that long having long duration storage would be somewhere in between where there is definitely a little more capacity needed to ensure reliability throughout those years but probably not as much as when there's no long duration storage right you just to make sure I understood you correctly here so you're saying that considering our resilience issues should actually tip the balance a bit more to long duration storage because that can come for free essentially exactly definitely yeah the seasonal seasonal role that long duration storage place is very important across the multiple weather years and here you are assuming that um the long duration storage technology could be more resilient than other baseload generation options that um is that correct I'm just trying to understand how you are how much yeah I sorry I think uh long duration storage is definitely more resilient than short duration storage but it's probably not as real resilient as a queen firm generating resource again because you need to rely on that extra generation for long duration storage because it's just the storage technology but the clean firm scenarios it's a generation source that's not as impacted by different weather years and so I think uh the long duration storage would lie somewhere in between I see so so the point is uh if we if we build peakers with CCS that should be the most reliable option uh in terms of providing resilience so uh long duration storage in other configuration would add to it but not necessarily reach the full extent of resilience that is possible so there's a discount that we have to take with long duration storage yeah yeah and that comes from it being an energy storage yeah maybe let me add one thing very interesting question here will and EJ um so this is probably under the assumption your long duration long duration storage output power is uh not high so for the extreme weather event let's say that Texas where you last that last for well let's see that it's really that several days uh to about weeks a week so you suddenly have this power needed going up uh so much uh if the long duration can handle high power as well then your long duration can coming in can output more power can you know satisfy that needs so this assumption behind I understand I just want to confirm well EJ right if that output power is high actually long duration storage could supply the extreme weather needs as well that's true and this is where I guess it gets interesting on how that long duration storage is being operated on a seasonal basis and it's like if that if that weather year happens to hit in the in the spring perhaps when you know there's a lot of uh there's not much stored in the long duration storage resource is there energy to pull out um but as we saw before you know typically at least in California the system was preparing for that you know summer peaks and holding the energy throughout the pole so it does seem like uh long duration energy storage would be you know sufficient if anything yes the additional kind of cost or resiliency need would come from having to outbuild that power capacity needs EJ and E I think this is a wonderful point um I think what we are getting to here is if long duration storage was more is efficient and can deliver high power that is really a game changer I think the quantification I'm personally looking for is you know if I increase the round trip efficiency by 10 percent or 20 percent I increase the power output by 2x or 3x how much should I pay more for that technology so you set five dollars to kilowatt hour as the target right so if you included those enhancements I would love to understand the increase in the value of long duration storage say you know if you can now use long duration storage a little bit for short duration then I can pay a little bit more for long duration and you know raising that cost floor can really help with adoption and technology development so I think that's one thing I didn't see in the analysis but that kind of cost modeling I think will be hugely valuable yeah and and just on that I let me share one more it's a little messy because and this is again not directly the point that was made as the costs are still a little off I think relative to what was what we're discussing here but this is the scenario where I mentioned where we were modeling a range of different power costs this is all at ten dollars per kilowatt hour and the different round trip efficiencies of long duration storage and so I think the most notable thing here is that it's actually not a linear kind of change in the value of the round trip efficiency so we actually see that for energy power costs when they're cheaper the round trip efficiency becomes a lot more valuable in reducing system cost because you see this is you know this decrease becomes a lot more steeper as we go and so per your question yes I can and even already scenarios where we were we already modeled this 85 percent scenario at five hundred dollars per kilowatt hour and we already saw in this case basically the corner one long duration storage was really the dominant storage resource and it really displaced a lot of the short duration storage as well and so you can imagine that if you are mad you know power costs don't even further down that this value might increase further as well so definitely the the point that you're making I think yeah ej this is a really great plus so in my what I'm looking at here is I'm cutting it horizontally across at a fixed systems cost say 75 percent relative to to baseline and then I'm looking at you know for example if you cut a 75 percent doubling the round trip efficiency from 40 to 80 means that you can pay four times more for the power cost I think that was my is that the correct interpretation of this plot if we cut it vertically like 75 so total the same system costs across all scenarios okay but then you look at the impact of the round trip efficiency on the power cost I'm willing to pay so I think am I right that doubling efficiency means you can you can pay four times higher in power right right so that's I think it'd be good to see if it's it looks like it's quite linear with the factor of two difference in the slope but I wonder if there will be a non-linear regime sort of that's what you're seeing toward the the low end of the power cost and that non-linearity could be a huge economical play I think to try to leverage that 100 percent and that's definitely what we see in the in the results as well and so yeah maybe we'll add another a couple more cost points on this curve there on the bottom at 100 and maybe 50 dollars and be happy to follow up with you on that will any but yes this is this kind of analysis really came out from the fact that we did recognize that there is a trade-off between round trip efficiency and power cost and we were trying to get to the dynamic there as well and it was very interesting to see the non-linearity outstanding E would you like to have the final question I think I'm all good so this is a great discussion here already I learned so much today EJ thank you yes I think we have a number of decision makers and policy experts in the audience and I definitely encourage those who are interested in learning more to read EJ's published work and also maybe to reach out to her directly to discuss further I think this is going to have far-reaching influences on future scenarios so I think it is still a long duration storage is still being sorted out right now and it will be the case for the next 15 to 20 years so I think your work is going to set the direction for the field very much so so EJ we really appreciate taking your early morning on Friday to speak to us and our listeners so I'd like to thank you once more and of course E thank you for co-hosting so Tracy or Evan if I can have the final set of slides please continue on the theme of long duration storage in two weeks we're going to hear from Dan Riker on other exciting long duration storage modes I invite everyone to tune in to hear from a very seasoned energy expert on the topic and this next talk will bring us to the conclusion of our summer series for the storage seminar and then we'll resume after that in October for another really exciting series as well and if I can have the next slide please and just as a reminder please do connect with us on social media you can also join our more in-depth technical talks that we call the storage of tech talks given by students postdoc and others in the Stanford community and then finally for those who are interested in learning more broadly about energy we have a professional education program that you can take classes from you can see some of the courses there we have a course on energy storage specifically so with that I'd like to thank everyone for tuning in this Friday morning thank you so much and have a great weekend