The Curtailment Paradox
All right, it's a few minutes past the hour, so let's go ahead and get started here. Thanks, everyone, for joining, and good morning from Colorado. I don't know what time zone everyone is in, but glad you all could join. And we're gonna be talking about the solar curtailment paradox. My name's Bethany Frew, and I'm an engineer at the National Renewable Energy Laboratory.
I've done work with capacity expansion modeling, production cost modeling, with a more recent in trying to couple those together to be able to explore wholesale electricity market design questions. That's kinda high-level areas of interest, but today's talk about curtailment is really looking at flexibility in the system and how, as we evolve to futures with high levels of PV and batteries, what sorts of interactions we see between different components of that system specifically related to curtailment. Make sure I get the slides here – there we go. So this talk is drawing from a paper that was recently published in Joule, "The curtailment paradox and the transition to high solar power systems." The link is on the screen. An amazing team of co-authors; many of them are on this call as well, and may help with questions at the end: Brian Sergi, Paul Denholm, Wesley Cole, Nathaniel Gates, Daniel Levie, and Robert Margolis. And as always, grateful for the funding from the US Department of Energy,
in this case the Solar Energies Technologies Office, who provided funding for this work. So what is a paradox? Some of you may know, but this actually came up as a question in our review process. The definition that I like and I think is most applicable to this particular paper and talk is "a seemingly absurd or self-contradictory statement or proposition that, when investigated or explained, may prove to be well founded or true." So basically something that at first doesn't really seem to make sense, but as you dig into it, you can start to explain what's going on underneath. And so in this paper we found two pieces of what we're calling the solar curtailment paradox. The first is that thermal generator parameters that relate to
flexibility, especially when we restrict minimum generation levels and ramp rates, impact variable renewable energy or VRE curtailment more in mid-PV contribution levels than at lower or at higher contribution levels. In our particular system, that mid-level was about 25 to 40 percent, and then the lower levels were about 20 percent, and higher, about 45. But those exact numbers may not translate to every system. The point is that there was sort of this wider band in the middle,
and I'll dig into this later in the talk. The second piece of this paradox is that allowing VRE and storage to provide operating reserves results in reduced operating cost and curtailment from a system-wide perspective, but the price suppression effect from these resources participating reduces the incentives for PV to provide operating reserves with curtailed energy. So in other words, this price suppression effect really kind of degrades any sort of monetary compensation that these resources might get for the benefits that they're providing from a system-wide operating cost and curtailment perspective. So I'll dig into each of these in a few minutes here, but I wanna give a little bit more context and background first before we do that deeper dive. So how did we get to this point with these two curtailment paradox pieces? We wanted to systematically explore the impact of flexibility options on curtailment levels as PV contributions grow. For example, how operational flexibility of thermal generators impacts curtailment; the role of battery storage; the role of transmission constraints, et cetera. And so we
assess the potential value also – and this is kind of that second curtailment paradox piece – the potential value of allowing VRE and storage to provide operating reserves with that curtailed energy. So basically let's understand how, from a very systematic perspective, various factors impact curtailment levels as we have more PV and batteries on the system; and then let's explore at least one way that we could try to use that curtailed energy for an additional value stream. Existing work in this space has looked at some of the factors that we explored, but not all of them in this sort of systematic comparison, and typically most of those studies were at much lower PV contribution levels, generally around 5 to 20 percent. There was one study, notably, that had
about 50 percent solar that explored the impact of minimum generation levels and start-up cost, but that was a more narrow kind of focus study. So why the focus on curtailment? Curtailment happens whenever generation exceeds demand and/or system conditions impose certain operational constraints that prevent available energy from being utilized. There's a lot of different factors that can play into this, but some of the sort of leading reasons are transmission congestion; minimum generation levels of thermal generators or hydropower, or back-feeding in the distribution system. And there's lots of literature out there if you're curious to learn more about kind of generally speaking curtailment. But we see curtailment really being a new normal in grid operations, and this is something that we've actually published on in the past, which is shown in the paper on this slide, where economic curtailment – and when I mean economic curtailment, I mean it's part of the least cost operations of the system, where curtailment is actually contributing flexibility to the system. It's really a tool that the operator has at its disposal. And so this is
really becoming a new normal in grid operations, again by helping to provide flexibility to ensure reliability. And in this previous work that I mentioned, we found many hours of very high levels of instantaneous curtailment, above 40 percent in these high PV penetration cases. So why the focus on PV? Many of the studies, both in the US and globally that are looking at sort of projecting or forecasting what deployment of various systems will be in the future, especially out to about 2050, are projecting for PV to be the largest renewable technology deployed out to that horizon. So basically we're anticipating a very large growth in PV deployment. And then also, because of the coincident nature of solar, where you have sort of that restricted sunrise and sunset element, there's a more rapid increase in curtailment as contribution levels increase for PV. At least PV alone; if you combine it with wind, it kinda dampens that effect, but wind doesn't see quite as sharp of a rise. So PV is a little bit more sensitive to curtailment, and we're expecting to see a lot of it in the system. So that's why we're looking at PV in curtailment,
but who cares? Who do we expect to try to take these results and do something with them? Our overarching question was really kind of broadly, from a very stylized manner, to understand or explore what challenges or interactions do grid stakeholders need to confront as PV contributions grow. So we see two main groups of these stakeholders that we think would benefit from these results. First is system planners and operators where thermal generator operations are particularly important at these mid-PV penetration levels – that's the 25 to 40 percent that I mentioned earlier, but again, may vary by system. And this suggests that there might need to be a phased approach as our power systems continue to transform and presumably move towards these higher VRE and PV contribution levels. So there might need to be sort of different regimes for both planning
and operations. And the second group are market designers, who might need to potentially explore revising operating reserve eligibility rules and compensation structures as PV contributions grow, and where there might need to be kind of resets or tweaks to be able to better align the benefits and value that resources are providing the system and the compensation that they're receiving. But I should say that's an even broader conversation in the market design world, and worthy of its own presentation. So our study system that we looked at in this particular analysis was based on the Los Angeles Department of Water and Power, or LADWP. You might've heard about the LA100 study
that recently came out from NREL. We were using basically one of the data sets that that team used with some slight modifications, and the link for the LA100 study is here. If you haven't seen it, you should go check it out. Some really great and really innovative work. So first we used a capacity expansion model to develop six PV and storage build-out cases where we assumed consistent load, both on the profile and the magnitude across those six cases.
We really wanted to have six different PV and storage contribution levels with all else kind of being uniform for those load levels, at least. And so within this capacity expansion modeling framework, it captures the declining capacity credit of variable renewable resources and storage resources, and so it inherently is endogenizing within its solution the need to over-build from a name-plate capacity perspective in order to have sufficient firm capacity for resource adequacy purposes. That was a question that I've gotten in the past, so just wanted to clarify that point. And also, because it's sort of this co-optimized framework, kind of a central planner perspective, it's inherently guaranteeing cost recovery for these six kind of base cases only. We then took those six different build-out cases and applied them to a production cost model. PLEXOS was our tool in this particular case. It's a commercial-grade software that some of you might be familiar with. And we did that to then explore various
sensitivities on top of those six base cases, which allowed us to explore those different operational factors and how they impact curtailment. So we looked at curtailment of course, generation costs, and price outcomes for each of those sensitivity cases as well as the base case. But because we've effectively modified the system from an operational standpoint – again, just those operational parameters; we didn't change any of the portfolio or the system build-out itself – we can't guarantee necessarily long-run cost recovery, because we're not accounting for that feedback effect on how those operational parameters would impact prices and, in turn, investment decisions, and even from a central planner perspective. So just wanted to highlight some of the kinda caveats and nuances of the modeling tools that we used here.
And then this picture on the right is showing kind of the broad footprint that was used for the capacity expansion modeling, which is the various different-colored circles. We used NREL's RPM, the regional planning model tool, for that to get those six different build-out cases. And so it looks at, in this case, the full western interconnection with sort of this more resolved kind of granularity from a spatial and operational perspective for just that LADWP footprint. But by having sort of a less resolved representation of the rest of the interconnect, it's able to account for the boundary effects. And so within that LADWP footprint, it's able to do mixed integer programming as well as – so that means unit commitment is represented, and that's done at an individual unit level, and individual transmission lines, as well. So slightly different from the ReEDS model, if any of you are familiar with that, so a little bit more granular, and it's meant to kinda zoom in, as the name implies, to a specific region. Okay, so these six build-out
cases that I mentioned, and this'll be consistent through the rest of this talk, we've named them by their annual PV contribution level for that base case. Now, of course when we've adjusted the operational parameters, the exact penetration levels or contribution levels change, and I'll get to that later. But you'll keep seeing these same six numbers throughout the talk, and that's just referring to these different build-out levels. Son installed capacity on the left, total generation on the right. The key point here is that you see solar contribution levels increase significantly
across these different cases, as well as battery contributions. And so the PV _____ note includes rooftop PV, utility-scale PV – and that's kinda this lighter color yellow – and then also we had hybrid PV-plus-battery resources, and we've separated in many cases at least the battery from the PV contribution of that overall hybrid system, just to kinda give a better idea of the breakdown. And so that's kinda the two different slightly gold colors in these figures. All right, so if we zoom into just, again, these six base cases without any of the other sensitivities, this table is just giving a little bit more information about what the contribution levels from the various resources are, kind of what's happening. I'm not gonna talk
through every number here. You can go look at the paper if you really wanna study this a little bit more. But the point is that we have a wide range of renewable contribution levels: if we just look at PV; if we look at PV plus wind; and if we look at sort of a broader renewable energy level, all the way up to a 44 percent PV contribution level with 96 percent instantaneous penetration cases, and then 75 percent if we account for wind. Storage ranges its contribution as a percent of peak load, so on a capacity basis all the way up to about 37 percent. And then
you'll see curtailment, the average annual ranges for total VRE about 8.2 percent in the maximum case and almost 10 percent if we look at just the PV resource alone. And then the max instantaneous, we're hitting 57 percent for wind and solar combined, and 65 percent for PV.
So the sensitivity scenarios that I've mentioned a few times now – again, we layered those on top of the base case, and so I know there's a lot of words on this slide with this table. I'm gonna kinda talk through the categories individually. Not gonna cover every single word here, but again, you can go to the paper if you wanna revisit this. So the base case is this first line,
so that's referring to those six build-out levels. And so this table is basically saying for each of these six build-out levels, we're going to apply an additional 13 sensitivity cases which in total results in 84 total scenarios that we ran and included in this report. So the first group here are what we're kind of lumping together as thermal plant flexibility. So this is where we have more and less flexible from a minimum generation perspective, from a minimum up and down time, and from a ramp rate perspective. So this is really trying to capture the ramp rate,
the duration, and the magnitude of these thermal flexibility parameters, to understand how they impact curtailment and specifically how much VRE variability and uncertainty the system can absorb. And so the duck curve I know is overused, but a lot of the challenges that folks point out from that figure are embedded within these sensitivity cases. So the belly of the duck, for example, is the minimum generation point. The next category here is really looking at those operating reserve eligibility rules, so that was kind of that second paradox element that we talked about earlier. So here, this is where we're allowing VRE and storage to provide operating reserves in different combinations of that eligibility, and basically to see if by allowing those resources to participate, if they can use otherwise curtailed energy in the upward direction, but also if they can curtail even further in the downward direction to contribute to those reserves, 'cause we have both directions.
And then other constraints, sort of a miscellaneous bag here, but looking at how the rest of system operations – for example unit commitment decisions resulting from each of these – can impact curtailment levels. And so we're looking at a greater resolution with the five-minute; we're looking at sort of the pass down from the day ahead to real time, with forecasting errors being seen when the real-time phase of the model runs. A no-storage case is really a counter-factual, and then copperplate to look at the impact of relaxed transmission constraints. So just a little bit more information on the operating reserve
scenarios before I dig into results. The eligible generators may provide operating reserves up to their operating limits, so just wanna clarify this in case there's any confusion. So that means that all of the ramping limits, the maximum resource availability – for example, you can't provide more reserves than the PV resource that you have available – and also online status constraints are all applied. So PLEXOS was able to consider all of these different factors to be sure that a resource actually would have had room to be able to provide reserves. And within PLEXOS we only modeled the sort of holding of the capacity; we don't actually model kind of the releasing or the deployment of that capacity. So that's a slight nuance to clarify as well.
But the point is that even with holding that capacity, its ensuring that all of these operational factors would be maintained if that capacity was released into energy. So there're six products. Not gonna go through all the details here, but just wanna highlight a little bit. We have regulation up and down; spinning; non-spinning; and flexibility up and down. Different response speeds for each of these resources, different scarcity pricing assumptions with sort of a hierarchy being maintained with those assumptions of the prices. Somewhat of an arbitrary choice here for how we just changed it by one dollar, but the point was do maintain that normal hierarchy that we see in the actual system operations. So we're gonna dig into the two different paradox components now. The first one was really that that
thermal flexibility only matters in the middle. Kind of a colloquial spin on the wording there. This figure, which is a key figure, shows VRE curtailment. So this is wind and solar, on the vertical axis; PV penetration on the horizontal axis; and the text boxes above the lines show the storage contribution levels, which if you remember were shown in that kind of summary table of those six different build-out cases that I showed earlier. And then the different lines here are the different sensitivity cases, with all the cases except for the ones that refer to operating reserves. We're gonna save that and focus on those when I discuss the second curtailment paradox item. So what you see here is that the no storage case, which is kind of that goldish-tan color on the far left – we see massive increase in curtailment as PV penetrations grow.
Again, this really serves as a counter-factual case. This would never really happen in reality, because pretty much all the PV being built these days has battery coupled with it, or at least paired with it as a co-located resource. Or hybrid, I guess there's a combination of both. But the point is that this was just showing the importance of having storage to minimize curtailment in the system. Kind of a bookend result here. And so the key point is that we have this transition zone in the middle, or if you wanna call it a Goldilocks zone, but it's at these mid-PV- contribution levels where those thermal generator parameters impact curtailment the most, much more than at the lower or higher levels. And this is where the Goldilocks term kinda comes in:
it's because in those mid-levels you have enough PV to kinda see curtailment, and you also still have enough thermal generators on the system for the flexibility parameters to impact curtailment. So as we start to get the higher PV contribution levels, for example, there's just not as much thermal online for those parameters to matter. So kind of that sweet spot, just right – the Goldilocks zone. And so we see significantly less curtailment impacts from that miscellaneous bag, that last column in the scenario table that I showed earlier. So the forecasting errors, the resolution, and the transmission – we saw significantly less curtailment impacts from those parameters, at least in this particular test system.
So thermal generator flexibility. I wanna kind of dig into those six different scenarios that are really looking at, well, the thermal generator flexibility, so kind of ignoring that miscellaneous set that I just described at the end of the previous slide. So this slide is showing the change in generation from the base case. So positive values here mean greater generation relative to the base case; negative means less generation relative to the base case. And I'm showing the six different thermal flexibility cases that have greater and less flexibility with sort of the green labels indicating greater, and the red showing less flexibility, and then we have kind of an aggregated view of the different generator technology types here. So key points are that in scenarios with greater flexibility, so the green labels, we generally see more generation from solar, gas CCs, and coal, with less generation from gas CT generators. In scenarios with less thermal generator flexibility, there's typically less
coal and solar generation, with slightly nuanced trade-offs among the natural-gas-fired generation. And what's really going on here, kinda the crux of all this, is that there's a trade-off between cost and flexibility of these different generators. So for example, coal tends to be relatively cheap but inflexible. Gas CTs, on the other hand, tend to be relatively expensive but very flexible, and gas CC is in the middle. And so as you restrict or relax different flexibility constraints for those thermal generators, it allows the system to sort of shift to whatever's the most economic, or what's available to meet the system needs. And so I'm gonna show an example in a minute here, but just wanted to note, too, that the minimum generation levels and the ramp rates, or at least the less-flexible case of the ramp rate, the 10 percent case, yielded the largest difference in generation.
So I'm gonna, for this example, to kind of talk about that trade-off of cost to flexibility. I'm gonna look at the zero min gen case, kinda that far left bar in this plot that I just highlighted in the red box. So in that particular case, at the 33 percent PV contribution level, this figure is showing a dispatch for four representative or kind of just example days, with the base case on the top row, and then that zero min gen case on the bottom row. So this is, again, a greater flexibility 'cause you're allowing those thermal generators to go all the way down to the minimum, and they don't have to stay at this sort of higher level when they're remaining online. The textboxes
in sort of the upper right of the upper right of these two different rows are showing the total generation cost and the curtailment for just those four days. So what we see here, again this is that trade-off in cost and flexibility, is that the additional flexibility afforded to the coal generators – because remember, they tend to be the most inflexible – so now they're pretty flexible, but by not having a min gen level, the system can utilize more of the zero margin of cost solar and this relatively low-cost but now sufficiently flexible coal, as well as the gas CC, which also has a fair amount of flexibility and is kind of the next tranche up in terms of cost. And so relative to the base case for these four days, we see a greater utilization of solar, we see quite significantly less curtailment, and also some operating cost savings, about 16.5 percent. So now I wanna move on to the second solar curtailment paradox piece, which – again kind of a colloquial phrasing here – is that curtailed PV eats its own lunch. At least with respect to operating reserves in this particular case, or this application. So what we see in this first figure, which is the same layout as that other curtailment figure that I showed before – so VRE curtailment on the vertical, PV penetration or contribution on the horizontal – now we're looking at just the cases that refer to operating reserves. So we see the base case still
with this thick black line, but we see the three different operating reserve sensitivity cases with the various orange lines. So the first thing I wanna point out here is that at high VRE contribution levels, by not allowing VRE and storage to provide reserves – so that's the no VRE or storage reserve scenario, which is the thin, dotted line, the upper one in this figure – we see significant increases in curtailment. And so the base case is the one where we allow both VRE and storage to provide reserves. So just to kinda make this super clear, I'm looking at this upper right point. And so this is the highest PV contribution level, where we see about 53 percent more curtailment relative to the base case when we restrict VRE and storage from providing reserves, so a massive increase. So let's look at what's happening at that particular point.
So this is the base scenario on the left column – the no VRE or storage reserves, which was that one that was 53 percent more curtailment, on the right-hand side here, and this is at the 44 percent PV contribution level, so that exact same point that I just highlighted in the previous slide. These figures are showing generation or committed capacity for the three different main kinda thermal units that impact the system the most: the gas CCs, the gas CTs, and in this case biopower. At least for this example, these three are the most important. The darker-colored lines in these figures show the available committed capacity, and the lighter-colored lines show the generation that's actually resulting from those generators. The text boxes that are kind of in the upper left of each of these panels is the generation. In this
case, it's for the full year, so the 8760 total annual generation for each of these individual generators. So what we see is that when VRE and storage can provide reserves, which is again the base scenario, thermal capacity that's otherwise committed to meet operating reserve requirements is no longer needed. And that's pretty clear in these figures, right? If you compare kind of the left case to the right, you see large portion of the year with this base case where we don't even need to commit these resources, and so therefore we also don't generate from them, because we're able to use that PV generation to help provide reserves. So this reduces generation for these thermal generators by about half, and you can see that from the numbers in these figures. And it also, as I just mentioned,
enables greater utilization of lower-cost VRE and storage resources not only to provide those reserves, but also for energy, thereby reducing curtailment. And that's why we see that 53 percent difference. So we thought this was a pretty interesting result, but first it seemed a little larger than we were expecting. But again, as we dive into this, it makes sense. Oh, and then it also reduces system-wide operating costs by about half as well, in this case. So some operating reserve price trends, just wanna touch on this quickly. Despite the curtailment
and system cost benefits when we allow VRE and storage to provide reserves, and the base case, this doesn't necessarily translate into increased revenues. So this is where there's kind of that potential misalignment in compensation that gets into some of the market design items. But higher curtailment levels associated with the increasing contribution of PV resources, as well as the interaction with storage – because we're assuming zero margin of cost for these resources, which is maybe another point to come back to, at least for storage – leads to lower prices for reserves, especially during the times of PV curtailment. So basically this is where it's eating its own lunch. Because they're low-cost resources, they effectively suppress prices and thereby prevent themselves from actually getting significant revenues by participating in these operating and reserve markets.
Operating reserve shortage conditions, and this is just kind of a secondary point here, generally only affect operating prices during non-curtailment periods in the transition zone. So, one, not really super prominent to begin with, and two, not very relevant for PV, but just kinda wanted to note that this was happening. It was just sort of a less-important dynamic that was happening. And this is the last set of figures I'm going to show. I wanna talk through a series of
price results to kinda help show what's happening with this price suppression effect that's kind of a high-level finding. What I'm gonna do is show kinda different panels here. Within each of those, it's the mean reserve price, and in this case looking at spinning reserve as the example. And so just kinda taking the average for three different kind of sets of hours across the year. So all hours of the year are kind of this dark plum/purple color; hours with no PV curtailment is this green or teal; and the only hours where PV curtailment is happening is the yellow color. And
then the six different PV contribution levels or build-out cases are shown for the different sets of bars. So what we see in this first case are no VRE or storage reserves. That's the one where we saw the most curtailment, the one that was 53 percent more than the base and that high PV contribution level. So what we see is that all three price metrics generally increase as PV contribution levels increase, so kind of an upward trend across all of these bars which is driven by the thermal generator commitments that we were just discussing. So you have higher-cost resources
that are being committed to meet reserves, and so they're the ones setting the prices for reserves. There's a much smaller secondary effect from the presence of operating reserve shortage or scarcity events during periods without PV curtailment. Those are the teal bars, so I kinda mentioned this already. It has the largest impact in that transition zone, but even with the sort of price impact, it has a very small impact on the overall system cost – less than 1 percent. But just kinda wanted to note that, not a super-strong trend happening. So now, if we move to the no storage reserves case, so now we allow VRE to provide reserves but not storage, we see reserve prices generally decreasing, because now curtailed VRE, which is zero marginal cost, can provide operating reserve capacity at the zero cost instead of the higher cost of those thermal generators.
And this is particularly prominent during periods of PV curtailment, which you can see by the yellow bars, and operating reserve shortage events play a much smaller role here. If we move to the bottom left here, where we now allow storage to provide reserves but not VRE resources, we see a much greater decrease in operating reserve prices which is driven by the participation of storage. Again, we're assuming a zero cost here, and it has ample capacity to provide reserves, especially during periods of curtailment, which are the yellow bars. Operating reserve shortage events play a very small role here. And then when we allow storage to provide reserves, we see a pretty significant reduction in total curtailment from a megawatt energy basis but also the number of hours where curtailment is happening, implying that there are fewer periods when PV could even use curtailed energy to provide reserves if it were permitted to do so. And in fact, that kind of thinking leads us into our final point here,
which is now going to our base case, where we have both VRE and storage allowed to provide reserves. We see the operating reserve prices don't really change much from this last case where only storage could provide reserves. Prices remain near zero, especially during times of curtailment. And so storage has really already taken most of the benefit of providing reserves, and so there's really little left for PV to benefit from that, kinda to that last slide and the last point, because there's just already fewer periods of curtailment in the first place. And so this highlights the importance of really understanding the role and interaction between PV and storage, but also for storage providing reserves. And the devil's in the details with a lot of these rules related to markets, and so there's maybe some of those details we need to explore to understand the implications or impacts that they could have on other resources in the rest of the system. So just wanna touch on a few wholesale market design considerations that kind of were relevant based on what we were seeing from this report. Again, this is more of a discussion and not
necessarily things that we are demonstrating in this study. Because the first caveat is that we're not actually modeling a formal wholesale market. As many of you probably know, LADWP is a vertically-integrated utility. It's not even really part of a formal market, although it certainly interacts with the surrounding _____. So really we're using sort of this production-cost
model, cost minimization approach to simulate at least from a high level how a potential market dynamic could play out, and how prices from those results would give us some sort of insight into how market design-related elements would be affected in these high PV contribution levels in the various sensitivities that we ran. So again, not a formal wholesale market, but the way that we modeled this gives us some idea of how prices might be affected to comment on where future work and future discussions should go with this topic. I didn't explain that well, but hopefully that made sense. So first, market structures and policies that allow all VRE resources to provide operating reserves yield lower-cost operations in curtailment. So this is really that second paradox item. But this sort of open participation is actually not common practice in competitive wholesale markets in the US, ERCOT being excluded from that, where they do allow basically any resource to participate as long as they can qualify.
Second, operating reserve markets may not provide a significant revenue source to compensate for reduced revenues resulting from greater levels of curtailment and declining energy prices. This is also really the second curtailment paradox piece. So really it's just saying that there might be a misalignment between the value that resources are providing the system and the compensation that they're receiving, and so there might need to be some adjustments to really compensate resources for the full value they're providing. And this extends, honestly, beyond even just the PV and battery discussion. This is sort of a broader grid system services
and getting the prices right within markets, and that's, I think, a topic that there's a lot of discussion and work being done on right now. But the third point is that markets may need to continually update pricing structures to better reflect value of reliability, flexibility, externalities, and the true lost opportunity cost. So again, this kind of is just playing on that same "getting the prices right" theme, where there's a lot of different rules and factors that can impact the price outcomes, and we wanna make sure that those price outcomes are appropriately signaling for the full set of grid services, and there's not any unintended consequences. So some of the factors that can play into this are the rules around shortage pricing, and so ORDCs – operating reserve demand curves – are a popular idea right now being implemented in multiple systems to try to better signal for that scarcity or shortage. Also bidding practices of different resources to make sure that they're bidding in to account for their full cost in the full set of considerations. So for example we assumed zero marginal cost for storage, but
really there's some sort of degradation cost since those batteries are being used – kinda this wear and tear idea, which extends to other resources as well. There can be participation rules that can affect bidding practices, and there could be other policy factors, like RPS – renewable portfolio standard – that may impact how certain resources need to be used, or how they did. There's also the avoided emissions pricing, so a lot of conversations right now about carbon pricing and carbon taxes. So that plays into how the prices turn out in these different markets. And also uplift payments, which are out of market, but they're used to really make sure that generators that are needed for reliability get compensated fairly. And so there's a whole conversation about that, and transparency is a big issue there. But the point is that there's a lot of details that play into how prices turn out, and trying to get those prices right is a big item to still be figured out within wholesale market design.
And the last point here is that markets may need to consider factors that impact how the rest of the system is operated, so min gen levels being an example of how different factors can impact operations. So for example, different emerging technologies like hybrids and distributed energy resources, and also resources that are effectively self-scheduled – for example if they are through some sort of bilateral contract or PTA where they may have flexibility that they can provide the system, but the operator may not see that or have access to it. And so there's lots of conversations about how to integrate resources to tap into the flexibility, or even just understand how the lack of that flexibility can impact operations in the rest of the system. Okay, so just to wrap up here, a few more kind of bullet lists. Sorry for all the words,
but I just wanna flag what we see as being some potential future work. Of course, this may not be exhaustive, but work that could look at differences in dimension, so different systems beyond just this LA-like model that we used, different build-out levels, different range of storage and hybrid contribution levels, different durations for storage. So lots of different cases where you look at a different build-out of the system and typology.
There could be cases that look at flexibility, both things that can impact flexibility but also supply flexibility needs to the system, as well as trade-offs between different options. So for example, and I think this may be the best example, is doing a full cost-benefit analysis of the thermal generator flexibility upgrades. So we basically just assumed, in our kind of bookend, stylized scenarios, that you would be able to have a zero min gen, for example. Is that possible? If it is, what's the cost? We didn't account for the feedback effect of that cost in the capacity expansion model, where it was making decisions to build and maintain certain generators in the system. So something where we try to co-optimize at a unit level to account for the full cost and benefit trade-off of these different flexibility options would be a really cool project to do in the future. Another one is looking at a larger role of demand response, or more broadly just price response of demand, particularly seeing how it could be an alternative to thermal generator upgrades. For example, could demand response result in the
same outcome as the zero min gen scenario? Also looking at consideration of other end uses for curtailed energy, so this gets into the ESI, energy systems integration, realm, where we're looking at kinda sector coupling and combining with other end uses beyond just the power sector. We could also look at more robust evaluation of the role of uncertainty, so this kind of gets into forecasting errors, and even using stochastic forecast, and the treatment for look-ahead horizons and resolutions, and how that sort of information flow, if you will, information flexibility in a sense, can impact operations as well. Also technology and market representation, so this kind of is a little bit of a repeat for some of the other points, but just looking at other types of resources and different ways that those resources can be operated, including the rules surrounding them. Particularly in this case, the storage and hybrid representation would be of prime importance. Also improvements to the operating reserve treatment, making sure that we're capturing the true cost of these resources and kind of the interaction with policies that may disincentivize, for example, down-direction of reserves, where there's some sort of RPS requirement. And finally, linking to investments – again, this also sort of touches on previous points as well, but kinda the feedback even more broadly from that cost-benefit co-optimization would be just more generally a feedback effect of operational sensitivities on investment decisions over multiple years of a system evolution. So
even just looking at market design impacts, for example, where there may not be a technology cost, but there's a feedback effect through the price signals. And also how market design changes could support revenue sufficiency for resources that are needed for reliability purposes. Okay, so I am done now. The reference is here if you wanna come back to those later, but thank you for your time. I know that was kinda maybe drinking from a firehose, but I will open this up to questions. And oh, I forgot to mention this at the beginning, sorry, but hopefully,
it looks like folks figured it out. There's a Q&A box, so please put your questions there, and I'm gonna go through those individually starting now, and may call on some of the co-authors who are on the phone to jump in if needed. So I'll just go in the order here. So, Joe Seale: "You mentioned cost recovery is only guaranteed in the base case. Can you comment on how PV revenue compares to LCOE with a large amount of curtailment and lowered wholesale prices at the times of solar generation?" Ooh, that's a great question. I don't know what the revenue versus LCOE breakdown is, but yeah, obviously with more curtailment, you're effectively increasing your levelized cost of energy from those resources because you're getting less utilization from them. We commented on this in the earlier paper – not this Joule paper, but the other one that I mentioned, from 2019 – but I don't remember the exact numbers off the top of my head. But I know it's in the text.
We didn't crunch these numbers for this study, so I can't give values there. But the point was just that there's sort of that feedback effect being accounted for in the base case, feedback with kinda prices within that capacity expansion model, but not in the sensitivity cases. So if that didn't answer your question, please feel free to add another follow-on. And if any of the co-authors have additional thoughts there, please jump in. Okay. "On slide 25, why do you say that storage has zero cost?" Let me go back to slide 25. Except I don't see – oh, there it is. I'm blocking my
– _____. Oh, I'm not saying that it actually has zero cost; I'm saying that's what we assumed, which really, y'know – Degradation cost, it's not actually zero, but it was a simplification that we made. But the point here is just that because we made that assumption, that's driving storage to really capture the bulk majority of the benefit from providing reserves. So future work definitely should explore what that actually cost for storage could be, and I know folks are working on that. It's kind of a can of worms, 'cause it gets hard to – How do
you parameterize that cost, like what is the cost curve? How do you track the usage of the battery to then align with that cost curve, assuming you're using a curve or function – whatever the math is behind it. But it's a tricky problem, and I know folks are looking into that. Okay, next question. "Did you include hydrogen as one of the sources – ? " I'm sorry, my screen just changed. "Did you include hydrogen as one of the storage options, in particular for longer-duration energy storage requirements at higher VRE?" No, we did not. We only included battery storage, as well as pumped hydro. And Brian, correct me if I'm wrong, we had a compressed air, I think, CAES. I can go back to that. We actually did have a small amount of demand response.
I guess I'm wrong about the case. Actually, let me go back to the – do-do-do – here we go. Oh, we do, okay, and we just didn't plot it in the other one 'cause it's not getting used. Okay, so there's a little bit of CAES, a tiny bit of demand response that you can see in the capacity. I guess the CAES and the demand response show up on the capacity but not on the total generation, so they're not actually being used. So pumped storage and battery are
really the two storage options that we used. But agree that in the future, looking at hydrogen, especially for long-duration storage, would be super interesting and, I think, valuable. Sorry, I'm getting my cursor tied up here. Okay, let me get back here. So Keith also asked, "On slide 30, one of the other end uses for curtailed energy is the whole H2 at scale wheel, such as the – yeah, exactly – steel, ammonia, yeah, transportation, et cetera. Yeah, totally agree, there's a ton of end uses where if you produce that hydrogen, even if it doesn't go back to the grid per se for balancing purposes, there's a whole bunch of industrial and commercial applications that can use it and be part of this kinda broader energy systems integration/ sector coupling. Yeah,
totally agree, and yeah, this would be an interesting area for future work. So the next question. "Did you consider a scenario with no coal generation, or where only the gas thermal plants could be run more flexibly?" Not explicitly in the sense that we did a no coal, but if you look at the – Actually, this is convenient 'cause it's shown right now. If you look at this
slide that's shown right now, you'll notice that at the two higher PV contribution levels, the 38 percent and 44 percent, there is no coal. Or, I guess, look at the installed capacity on the left-hand side. So the coal is all retired by that point, and this is part of the LA100 trying to get to 100 percent system. They had certain retirement decisions within the model.
So, yeah, so there was no coal in those cases. And this has actually come up before – one of the interesting things to do would maybe be to go back and look at an hour-by-hour basis from some of these runs, and break down by technology type sort of what's running versus what's not, to see like capacity factor values of the generators, to see maybe if we could kind of back out whether coal or gas is contributing more or less to the curtailment. That's not something we've had time to do, but there could be some just even post processing with these that could get at kinda the relative role of gas versus coal. Okay.
So a question from Yang. "If I understand correctly, this is based on a regional grid. Would you mind making some comments about whether it is technically feasible to scale this study up at a national level?" Sorry, my thing keeps moving. "…at a national level, or even at a global scale? Are conclusions here applicable to other regions and/or grid conditions?" Yeah, great question, and sorry if I didn't make this clear. Yes. These results we can only guarantee for this particular system, and that was sorta why we had flagged in future work looking at other systems and other configurations. It would be, I mean, technically feasible, 'cause there's other studies that have looked at the full US in a production cost-modeling perspective. It gets challenging to
run those. They take days with parallelization techniques on our fancy HBC high-performance computer. So it would be possible, but for the budget and time frame that we had in this study, doing those scale of runs was not possible. But certainly it could be done. I think it could be
interesting, especially where you start looking particularly in the Eastern Interconnection, where you have larger loads, and you have a more kind of interconnected and meshed grid. Seeing how the imports and exports from different regions – basically that transmission question, how that can play into this. Because yeah, we only looked at sort of the small footprint in the LA. So even though the capacity expansion model looked at the full region, the full Western Interconnection, we only looked at the LADWP footprint for the production cost modeling runs, effectively treating it as an island. Which may seem like a strong assumption, but it's actually consistent with how LADWP does their planning studies – basically kinda worst case, "Let's assume it's an islanded system." So it just made it easier to kind of maintain tractability
and dig into what was going on. But yeah, future work should definitely explore other systems. And we try to caveat in the report that some of the exact numbers and the quantitative results may not extend exactly. But yeah, future work should explore that. Okay, I'm not seeing any other questions on here, but let me just scroll through again.
I guess if no other questions pop in in the next minute or so, we can just end early and give folks a few minutes before what is likely their next meeting. And of course, we're always happy to answer questions by email, if you wanna follow up later, and I can put my last slide with my email on here in case you wanna note that. Here we go. Oh, I think I saw a question come in. "Did you look at individual impact on thermal plant profitability?" No, we did not. We could, or at least a net revenue perspective. This is what's tricky, and at least in the past work I've done with production cost modeling runs, net revenues is really the best we can do. Net revenues are
the total revenues from kind of the energy and ancillary services, their operating reserves in this case, minus the operating cost – so basically the production cost within the PLEXOS runs – and so that difference is called the net revenue. And that's actually used by the independent market monitors when they're assessing the various ISO and _____ regions for market design considerations, and market power, and that sort of thing. That revenue basically indicates what's left to cover the fixed costs, and so often that's compared to like a net CONE value, for example, just to determine whether there's sufficient net revenues to invest in new resources. So profitability gets a little tricky to report, but we could if we wanted to, and if we
had time and resources, could go back and calculate the net revenues for those generators. Great, so I guess maybe I'll just give it a few more seconds if there's any more lingering questions. Oh, now I'm seeing – Oh, there're several questions. Okay, also a follow-on. "Do you think that current market design broadly speaking incentivizes
updates of existing combined cycle plants at lower min load?" Oh, geez. I mean, there's a lot here. I mean, this kinda gets into the flexibility market design topics, and all the talk of adding flex products, flex ramping _____, for example. I mean, I think the idea is for those to signal for more flexible capabilities of resources. I can't comment on whether they're sufficient or not. I guess it just depends on the system. But I mean, I think that's the direction some of the market designers are doing with, and I mean and certainly some regions have already implemented it. Whether it's getting the right press or not is another question. But yeah, this is part of, also,
that full kind of co-optimization with the cost benefit that we'd wanna account for the cost to do those upgrades versus the revenues. "Does the model take into account scarcity pricing?" Yes, it does. So there was one slide, I kinda went over it quickly, but it showed the different scarcity pricing assumptions that we used for the operating reserves, which tend to be the more important set versus energy, for example, because you're gonna violate your reserve requirements, which leads to scarcity, before you ever drop load. So those scarcity pricing values are what you're gonna actually see. Oh, I think I have it much further back.
Oh, here it is. Yep, so here's the shortage or scarcity pricing values. Again, they're somewhat arbitrarily chosen just to have a hierarchy, where our regulation reserves tend to be at the top of the "pyramid" for our operating reserve hierarchy. So yeah, and we did see scarcity pricing, or shortage pricing, events that were driven by having a shortage in the reserve requirement – or meeting the reserve requirement. "Has the transmission constraint
have been considered in case of excess solar _____ generation?" Not sure I understand – So I'm not sure I understand the question, but I think what this is asking is about the transmission treatment. And in every scenario except this bottom and this figure, the one that says "copperplate," that's assuming transmission are not enforced, so a copperplate system, where there's no flow limits or anything. Aside from that case, every other scenario that we ran enforced transmission limits within the footprint that we modeled. Okay. Unless I've missed any, I think we've gotten through all the questions here. Give it a few more seconds here, and then if not, we'll wrap up. We're almost out of time anyhow.
Bah, okay. Oh, I see. Got it. Okay, thanks, Brian. Thank you for that. Cool. Well, I'm gonna call it. Like I said, we're almost out of time. But thank you all for joining. This was actually the first
time I've done one of these NREL webinars, so I'm excited to be able to share this work with you. Hopefully, if you're interested, you'll be able to go and look at the full report.