2022 Standard Scenarios and Cambium Data Set Release (Text Version)
This is the text version of the video 2022 Standard Scenarios and Cambium Data Set Release.
This video is a webinar that discusses the 2022 release of NREL’s Standard Scenarios, a set of annual scenarios for how the U.S. electricity sector could evolve through 2050.
>>Pieter Gagnon: All righty. I'm going to go ahead and get started. Thanks for joining us today. I'm Pieter Gagnon and I've got Wesley Cole and my colleagues on the call to help answer questions and try to have a little bit of dialogue at the end of this. There's also some names on this screen. We're going to talk today about the release of NREL's 2022 Standard Scenarios analysis and data set. I've got some other names on here. This is a big annual effort and it was particularly—a lot of work went into this year because of the passage of the Inflation Reduction Act and getting that into the model and a lot of exciting stuff. So, we're going to go ahead and get started.
The agenda—first going to—a brief introduction. What is the Standard Scenarios? What's the Scenario Suite? Stuff like that. Then we're going to talk a little bit, not too much detail but a little bit about the representation of the Inflation Reduction Act, the electricity sector provisions in it in this year's analysis. Then I'm going to talk a little bit about—I'm going to show some highlights from this year's Standard Scenario set. And then, at the end I'm going to talk a little bit about the Cambium data sets, which are a kind of an elaboration of the Standard Scenarios, and open it up for Q&A. If you do have questions, put them into the Q&A box. There's a little button thing that you can open up. Wesley might be able to answer some of those in real time; otherwise, we'll get to them at the end. And the webinar is being recorded and it will be posted online if you want to share it or reference anything.
What Are the Standard Scenarios?
So, NREL's Standard Scenarios. This is the eighth edition of our annual projections of the possible evolution of the U.S. electric sector across a wide range of possible futures. So, we have a big suite of scenarios that we model looking forward out through 2050. The workhorse here is the ReEDS model. It's a linear program and optimization model. It has a bunch of inputs into it. Some of them are other models, things like our dGen model, which tries to project the adoption of behind-the-meter solar, things like that, and then a bunch of different data sources about different resource availability, future fuel prices, technology, cost, and performance, etc. All of that goes into ReEDS, which then solves for the least-cost operation and building out the power sector subject to policy and operational constraints. And then, it outputs these mixtures of what gets built and how is the grid being operated in order to, again, operate at least cost. So, whenever we're showing you stuff about the future, that's what we're doing: least cost subject to policy and operational constraints with a linear optimization model.
The—I'm going to show you some results today, but you can also—there's several different products that you can go to. One of them is—there's a report that's out now. You can just go and Google "Standard Scenarios 2022 Report" and that kind of is a highlight and explains some of the major findings as well as the input assumptions into the report—sorry, into the analysis. There's also—we have this online scenario viewer and data downloader. So, that is here at scenarioviewer.nrel.gov. If you go to the Standard Scenarios landing page, you will get directed towards this. And you can go here to the Standard Scenarios and you can either view or download the data. I'm not really going to go into it here; I'm just letting you know that this exists. And it's kind of an interactive workspace. You can plot different metrics and different scenarios and compare things as well as just downloading CSVs if you want to then go and kind of plug in the data for your own plotting or analysis purposes.
So, both the report and the scenario viewer. And then, again, I'm not going to talk too much about this right now, but there is this Cambium data set where we take a subset of the scenarios and create more metrics at hourly resolution and then also make those useful through the scenario viewer. That's not out yet. It will be out very shortly, either this week or next week. But it is a kind of an elaboration for the subset of the scenarios. Anyway.
Okay. So, this year's Scenario Suite. Different—we always kind of tweak things as we go along. We start with a set of mid-case assumptions, mid-case meaning central estimates for things like technology costs, fuel prices, resource availability, kind of central estimate for demand growth, and we have existing policies as of September 2022. So, that does include the Inflation Reduction Act as well as whatever state and local policies existed at that point in time. So, this is kind of our business-as-usual starting point. It's not necessarily the thing that we think is most likely to happen; it's just central estimates for everything going forward and existing policy. So, that's what we start with.
Then, we have a handful of different sensitivities where we will take those assumptions, keep everything constant, but then change just a couple of things—so, things like low renewable energy and battery cost sensitivity, high renewable energy and battery cost sensitivity. Things like reducing renewable energy resource availability, low and high demand growth, low and high natural gas prices, as well as a scenario that has—does not have the Inflation Reduction Act in it so you can kind of compare and see how things changed because of IRA. So, we have these sensitivities on top of our mid-case.
And then, we have a couple other convolutions. That's not the right word. Things that we expand the set. The first one is that we have for each of these different sensitivities and the mid-case we run it for both an expansive technology set that includes a bunch of technologies that are still relatively nascent, things like enhanced geothermal systems, floating offshore wind, CCS technologies, small module reactors, and renewable energy combustion generators—that's things like hydrogen generators. But then, we also run all the scenarios without those nascent technologies, only having the things that are at this point more established. So, we have both sets for each of these different sensitivities.
And then, the last dimension to this Scenario Suite is that we run all of this, the sensitivities and the technology sets, with just policies as they exist today, but then we also have versions where we run the—project out the evolution of the grid subject to two different decarbonization trajectories. One of them has 95% reduction in electric sector CO2 emissions by 2050 and then another one is a lot more aggressive: It's 100% reduction by 2035.
So, combining all these things together, you end up with 70 different scenarios. So, you can go and download, look at all 70. I'm only going to be showing you a subset of them today.
Inflation Reduction Act
So, just very quickly— I'm trying to get to the results; that's the more interesting part, but I'm going to talk a little bit about the Inflation Reduction Act. And I'm not going to talk about everything here. The report has much more detail of exactly how we implemented it. One thing I want to be clear is that this is only a partial representation of the act. We implemented the main electricity sector provisions, but there's a lot of stuff like manufacturing incentives, a bunch of financial mechanisms, and stuff like that that might ultimately end up having an influence on the power sector that we did not at this first pass make an attempt to try to understand how they might influence the power sector. We really focused on the main electricity sector provisions of those.
The main ones that I'm going to briefly talk about here are four different categories of incentives. There's the production tax credit, the PTC. So, that's a per-megawatt-hour credit that generators can get for ten years. We assigned that to onshore wind, utility-scale PV, and biopower. Technologies can select between the PTC and the ITC. We just did a priority analysis to determine which one was likely to be more profitable for different technologies and assigned them, which is why they're labeled as such here. And this here, this note about the PTC is actually quite important. You'll be seeing the effects of this in the results, which is that the PTC will now under IRA phase out when electric sector emissions fall below 75% of 2022 levels—sorry, that should be "fall below 25% of 2022 levels, or a 75% reduction from 2022 levels"—or 2032, whichever is later. So, if we reach that threshold before 2032, the credit will stick around until 2032.
The ITC, we have it at 40% of installed costs. One thing I didn't mention is that there's some assumptions about bonus credits going into this. I don't have time to get into that. If we want to have a discussion about that at the end, if someone asks a question, then we can try to circle back if we have time. So, the ITC, 40% of installed costs assigned to these technologies here. And like the PTC, it phases out when emissions fall below that threshold.
Then, there's two other incentives. One is for captured CO2. There's different levels for CCS and direct air capture. That just expires in 2032. There's no dynamic phaseout of that credit. And then, the last thing is there is a credit for existing nuclear technologies. It's this thing—there's a credit level that is reduced if wholesale prices go above certain levels. We didn't try to represent that. We just made it so that nuclear is not subject to economic retirements in our modeling through 2032, which is how long the credit lasts.
There's other things that I'm not going to go into right now. Assumptions about debt and demand growth. We assumed that IRA would induce kind of light electrification—so, not aggressive electrification, but it would increase demand growth slightly. There's things about safe harbor provisions, distributed PV, financial stuff, etc. We can kick that to the report.
And then, the last thing that I'll mention is that we will have at some point in the next couple of months putting out an IRA-specific— an analysis specifically focused on IRA. That's not out yet. The Standard Scenarios has IRA in it. It's not really designed to really understand the effects of that policy, but you can get a lot out of it. So, if you really care about IRA, look. There will be a report coming out within a couple of months hopefully.
Highlights from the Standard Scenarios
Okay. All right. Thanks for hanging with that. I'm going to show the results now, some highlights from the results, which is certainly the more fun part. So, what I'm showing here, what we're going to start with is this mid-case scenario. This is that scenario that has the central estimates for things like technology costs, fuel prices, etc., and it has existing policies. This set that we're looking at here includes nascent technology, so it has things like CCS in it. I'm going to walk through what we're seeing here, starting with the bottom. On the left we have the generation mixture by technology type over time through 2050, and on the right we have the capacity mixture—so, how much gigawatts are built versus terawatt-hours of actual generation.
So, starting at the bottom, in our central projections we see fairly steady generation from the nuclear fleet. If we didn't have the IRA PTC, then we would see that declining, but with the IRA existing generator, nuclear generator PTC they stick around and they kind of make it through a tough spot and then hang out through the end of the horizon. But we don't see new stuff, new nukes being built.
Coal really declines quite a bit through the 2020s but then hangs around a bit at relatively low generation levels, being used in relatively low capacity factors throughout the kind of rest of the decades.
This little gray sliver here is coal CCS. So, we actually do see some retrofits occurring of coal plants. This gray—I'm getting a little bit ahead of myself—this gray, this lighter gray bit here is natural gas CCS. One thing I do want to be clear is the actual—what we're seeing here is retrofits of fossil units into CCS units being induced because of the rather large incentive for captured carbon under IRA. There is—at this point we just had a single retrofit cost assumption for different generators. In reality, it's going to be a lot more complex. Some generators are going to be easier to retrofit. Some would be extremely expensive to retrofit. It seems pretty clear that there are going to be at least some generators that are in the money for doing CCS retrofits; however, this is definitely an area that is going to need much more detailed analysis probably at a plant level to really understand kind of the future of CCS, especially CCS retrofits.
So, what we have here, we're seeing adoption. Seems like some things are in the money for at least some generators, but you really—if you care about CCS, you should look to forthcoming work, more detailed work about kind of generator-specific upgrade costs. Anyway, forgive that digression.
To continue looking at the stack here, we see natural gas remaining fairly constant throughout this horizon. We have a handful of technologies here, things like hydro, Canadian inputs, geothermal, biopower. I'm not going to talk too much about those because they remain fairly constant. Which gets us to the last two big ones. The blue is wind and the yellow is PV. Those are really where a lot of the growth is happening in the mid-case because of IRA, because of the significant incentives.
Now, you may already have noticed and wondered what is going on here in these kind of later time periods, because you have gas kind of starting to expand and actually a little bit of a tightening of wind. That is the expiration of IRA's PTC and ITC for renewable energy technologies. The threshold in these scenarios is actually crossed sometime in the mid-2030s and then the technology—those credits phase out. There's a safe harbor period and a phase-down period, so by the early 2040s you don't have those credits anymore, which means you really largely stop building new renewables and most of the new growth is served by the increase and utilization of the existing gas fleet.
So, this is important when we look out far ahead and we look past the expiration of the IRA tax credits. You can see this kind of rebound or reversal of trends in our central projections. But of course, you should take that—you should understand that with a certain amount of humility because that is of course in the 2040s. That's a long ways away. A lot of things are going to happen by them. Whether or not the tax credits are actually going to phase out is always a—it's a policy question that we don't have any more insight than anyone else on. I will say that PTC and ITC for renewable technologies has been scheduled to phase out at many points in the past and we've always renewed them, which is one of the reasons why we have sets of scenarios where we don't have it phase out and we just have it continue. But you're going to see this phenomenon present in a lot of the scenarios, this kind of rebound in later years when the tax credits phase out, if they do. Some scenarios don't reach the threshold.
Always fun to compare against what would have—what our projections would be if the Inflation Reduction Act was not passed. That's what I've got here. The top one is the technology—is the projection, the mid-case projection that I was just showing you with IRA and the bottom one is without IRA. So, it's pre-IRA policies as of September 2022—so, the federal policies that existed at that point in the absence of IRA. So, you can see a big thing, the big first thing that you notice is a lot less wind and solar. There's a little bit of growth, but not like the—what we expect under IRA. "Expect" is a strong word. What is the modeled projection of the cost-optimal solution under IRA. There was a little bit more growth in gas, and coal stuck around a lot more. So, you can really get a sense of how impactful IRA was by comparing these two results.
Another thing that I just want to show is, again, at the top here, just for reference, we have the mid-case with current policies, but then down here I'm showing the mid-case, everything the same except we put on a constraint to have the power sector—the US electric sector, I should say, achieve 100% net decarbonization by 2035. Here you can start to see some other technologies beyond just CCS that we consider to be nascent technologies show up. Those are—it's kind of hard to see, but there is renewable energy combustion turbines being built out, mostly provide from capacity, And then, we do have a little bit of bioenergy with CCS—that's kind of the neon green here—which is in our counting considered a net CO2 removal technology, so it's being used to kind of offset some gas technologies that stick around in this scenario to achieve net zero decarbonization.
So, one thing that you see here is that in our central scenario an absence of decarbonization policies—or an absence of additional policies. We don't really see some of these renewable energy combustion turbines or bioenergy with CCS show up. But if you do, you can fairly easily get to the point. You can—I shouldn't say easily. That doesn't mean anything. You can impose conditions on the model where these more nascent technologies start to play a fairly important role kind of solving the last little difficult part of decarbonization.
I'm going to walk a little bit now through—this is—these are plots of the ranges of generation from now through 2050 by different technology classes. And the main thing that I want to do here is just show you—these are—each one of these has 70 different lines, one for each of the scenarios that we ran. And this is really helpful to remember the significant uncertainty that we're looking at here. And this isn't even – we're not trying to pretend that this is an exhaustive description of the possible futures of the US electric sector. There's many things that could happen that would fall outside of these ranges. So, I'm going to walk through the different technologies one by one here quickly and just kind of highlight some of the more interesting aspects of them.
Starting with wind and solar, as you saw with the mid-case, this is where a lot of the growth is happening. Generation levels from these two technologies increased significantly across all scenarios in—for both of these technologies. And my apologies, I just realized I didn't mention what the line colors are. We have—these kind of brownish lines are the "no IRA" policies. The yellow lines are all the scenarios with "just current" policies. All the green lines have the 95% decarbonization trajectory. And all the blue lines have the 100% decarbonization trajectory.
So, you can see here that for solar and wind there's—not only is there significant growth, a pretty big range where we can end up with in 2050 but it's—also, they're all materially above what growth we projected in absence of IRA. And then, you can also see, as we pointed out before, that especially in the current policy scenarios you see a leveling off of solar and actually a slight decline in wind. That's a decline from retirements that are not repowered once the tax credits phase out in the scenarios that they do. But as you can see here, not all of them—some of them, there's decarbonization policies that keep growth happening; other ones don't expire. The threshold is never met, so the tax credits never expire. A whole bunch of things can happen.
Natural gas, another big range. You can see here this kind of heavy, dashed, yellow line is what we were looking at before, where you can see kind of a decrease in natural gas generation but then a rebound once the reversal, once the tax credits phase out at that mid-case scenario. Pretty much everything—yeah, everything is below where the pre-IRA natural gas levels were. But in most scenarios, most current policy scenarios there's still quite a significant amount of natural gas generation through the modeling horizon with different stories being told for the decarbonization scenarios. In no scenarios do we see natural gas going away. You can see it kind of leveling out here at 250 terawatt-hours per year even in the 100% by 2035 scenario, which is being offset by negative emission technologies. So, yeah, we have the rebounding here.
Coal really goes down a lot in generation. Again, a lot of that capacity tends to stick around in many of these scenarios but the capacity factors are a lot less, way lower than the pre-IRA coal projections.
Nuclear, most of the scenarios it sticks around. Some of them, we see some retirements. In a couple scenarios, we see new nuclear builds, but not many.
And then, the last thing is, again, just pointing out that we do see CCS builds in our scenarios, including some biopower builds when you need that—or, when it's beneficial or helpful to have that negative emissions technology offsetting uncontrolled—or even offsetting things like natural gas CCS technologies in the 100% net scenarios. Again, I just want to emphasize that CCS is particularly subject to—I'm positive that our understanding of CCS and its role in the power sector is going to evolve significantly over the coming years, so you should pay attention to new research in that area that will be coming out soon.
Okay. This is one of the last plots of the highlights. This is just electric sector—US electric sector CO2 emissions from fuel combustion for electricity generation in million metric tons per year across the 70 different scenarios. This dashed line here is that IRA emissions threshold. So, whenever a scenario crosses that line the IRA tax credits would start to phase out. Again, there's a safe harbor period and the phaseout period is not immediate, so you don't see a rebound right away. So, here in this—the heavy, dashed, yellow line you can see it crossing the line in, I guess, the early 2030s, but then it takes until the late 2030s to actually have the reversals start to happen because generators are still capturing the incentive for a little while afterwards. Anyway.
The main thing that you can see here is (a) a significant decrease across all scenarios in the coming decade, and then (b) a lot of scenarios cross this threshold, which would then induce the tax credit phaseout. However, not all of them do. Things like if we have significant electrification, those scenarios tend to kind of get close to the line but they don't cross it until maybe very at the end, or they don't cross it at all, or other scenarios just don't end up crossing these lines and therefore don't see the tax credits phase out. They just persist indefinitely in our modeling horizon. Yeah. And then, of course, the decarb scenarios have at some point a binding trajectory for either 95% reduction or 100% reduction. And then, pre-IRA is this brown line. Again, you can see how it was declining but not as rapidly as the kind of new outlook for the power sector.
Okay. Just a reminder for where you can get everything we're talking about. One, there's the report, Standard Scenarios 2022 NREL should get you there. I should have a link on here but I don't. We have the scenario viewer and data downloader if you actually want to play with the data yourself. And then, we will have forthcoming the Cambium data sets. If you just Google—there will be an email announcement. If you Google "Cambium NREL" you will get to a landing page which will update when the data is out, as well as you can get the Cambium data through this same scenario viewer and downloader if you go here. You can see previous iterations of Cambium are here as well as previous iterations of the Standard Scenarios.
Cambium Data Sets
Okay. I have just a couple things that I'm going to say about the Cambium data sets. Because they're not out yet, I'm not going to be showing a lot of things. But I did want to just give a little bit of sense of what's in here, highlight one of the important outcomes of this and, yeah, just talk very briefly about it. So, the first thing is the Cambium data sets, we've been doing Standard Scenarios for eight years and then this will be the third edition of the Cambium data sets, where we take a subset of the Standard Scenarios and then run them through production cost modeling to get more metrics at hourly resolution. So, the Standard Scenario is stuff like this, the things we've been looking at, these kind of annual plots of generation mixtures, capacity, stuff like that. Cambium data, tons more stuff there. There's things like marginal cost metrics, operational metrics, like what's the hourly mixture of generation for different parts of the country. But then, there's also things like emission metrics and other stuff like that, which is really designed to support other analysis and decision making.
So, Standard Scenarios is really a—you can really use that to—you can understand future mixtures that we're projecting but it's really kind of narrative. It's like trying to understand the potential future—have a conceptual understanding of the potential future of the power sector. The Cambium data is really meant to be something that you can actually take and plug into your analysis for trying to say, "What are the emissions induced by—if we built out a bunch of electrolyzers?" Sorry.
I'm not going to go into detail here. Like I said, there's a subset. We don't do all 70 scenarios in the Standard Scenarios in Cambium. We're going to have ten this year. That's up from five last year. So, we have the mid-case and then we have perturbations of low renewable energy costs, high renewable energy costs. We are going to have an electrification scenario this year with greater demand growth, low and high natural gas prices, and then two decarbonization scenarios. So, I'm not going to go into the detail here but we do have a suite of different futures with this additional data.
I'm not going to go into the detail here either, but let's just say there's lots of different metrics. There's a bunch of operational metrics, things like the generation mixtures, transmission and load, stuff like that. There's marginal cost metrics so you can see how energy and capacity costs evolve over time.
But then, really, what gets a lot of attention and engagement is we have various metrics related to greenhouse gas emissions. And one of the reasons for that is that there is a metric here, the long-run marginal emission rate, which is both particularly useful and there's not another national data set where you can go and download it. There's other groups that are producing these types of data for specific states. The consulting firm E3 produces long-run marginal emission rates for California and New York, for example, and there's other groups that do it for international countries and stuff like that. But if you want to try to understand the avoided or induced emissions from electric sector interventions, this is a metric that's designed for that and we get a lot of engagement because there is a lot of—if you are building a building and you want to understand the emissions consequences of different design decisions, there's a lot of change happening in the power sector and it's helpful to look forward and have metrics that are designed specifically for that purpose.
I'm just going to define this real quickly. I probably should have done that before talking about it. But this is—out of all of the metrics that we produce in Cambium, this is the one that is most—it's useful for many different types of analyses if you care about greenhouse gas emissions, induced or avoided. What is it? It's an estimate of the rate of change of emissions from electric sector intervention, taking into account both the operational and structural effects of your change. And that structural part is the key thing here. Many of you, if you just heard the term "marginal emissions rates," it's almost certainly that what the person is talking about is the short-run margin that is just an operational metric. It's just saying: taking the grid as it is right now, if you ask for another unit of power, what is the emissions induced from the grid as it exists at this point? However, if you are trying to say, for example, "What if we electrify the municipal vehicle fleet of Denver?" or build a bunch of electrolyzers or something like that, that is going to be a long-term change in the electric sector demand, and in most scenarios it would be reasonable to assume that we will build new things in response to that change in load. And this metric takes that into account. So, if you add a bunch of electric vehicles and we think that you're going to build some wind and solar and gas generators in response to that new load, this metric folds that in and combines both the operational and structural consequences of the intervention that you're analyzing: energy efficiency measures, electric vehicle loads, geothermal heat pumps, whatever. And we mean that term expansively.
One of the main things is that the numbers, this question of what is the induced emissions from an increase in load or avoided emissions from a decrease in load, are—tend to be relatively low, certainly lower than the short-run marginal emission rate. This number, this heat map right here is the kilograms of CO2e per megawatt-hour for the nation as a whole over a 20-year period in our mid-case. And you can see here that especially there's this relatively strong diurnal trend where midday loads, because they're expected to induce—or, we expect them to induce more solar builds, tend to have what end up being quite low numbers. I'm guessing a lot of people don't have an intuition about this, but these are really quite low numbers for induced emission rates, and even kind of the off-hours are —still have a lot of wind and other stuff in there, so it tends to be quite low. So, this is one of the main things that is in the Cambium data set and we get a lot of engagement on that will be coming out shortly.
If you care about anything that I just talked about—long-run marginal emission rates, how they're different from short-run, what do you want to use when—we do have a paper that we put out recently really talking about these different metrics and kind of explaining where one might want to use one or the other and why they work in some situations and don't, you can find that here: "Planning for the evolution of the electric grid with a long-run marginal emission rate." If you care about that, I encourage you to go check out that paper.
And that's what I got. So, Wes, you—do you have some questions?
>>Wesley Cole: Yes, quite a few actually to pass over. So, I want to start with this one, just as people have been going to the scenario viewer and looking at what's in there and they're noticing that especially the storage results are different in the scenario viewer than what you've been showing here for generation, that it's not showing up in your stack here but it's showing up in the stack in the viewer. Do you want to talk about differences—?
>>Pieter Gagnon: Yeah. Yeah. Sorry, we're kind of inconsistent about this. I'm assuming that you're talking about why is storage not showing up in the generation plots? Apologies. We are sometimes inconsistent with how we show storage. Sometimes we show it as a net generation decrease—or it's the net impact, so it's a decrease. Sometimes we just plot the generation from storage. Sometimes we don't plot it at all. That's what we had here. So, sorry, that was just an omission. It should probably—I prefer it the other way—so, the way that it's in the viewer where we actually show the storage contribution to the generation. The thing is, it's a little bit confusing because of course that storage is taking from other stuff. Anyway. So, that was just an inconsistency in how we're plotting things. You can see storage is being—this is storage capacity on the right, so we have quite a bit of storage and it is playing a role. If you go and download the data, there will be generation values for storage. I probably should have plotted it here.
>>Wesley Cole: So, as summary, it should—the data should match between what you see here and what you see there. They just may be different ways of presenting it, I guess, is the piece there. So, we've had quite a few questions about concentrating solar power and geothermal. It's—if you look at these plots, it's like you've got something for CSP on there and you've got geothermal on there, but you can't see them. Can you talk toward the potential of those technologies that—smaller in today's generation mix, get these—they're under the IRA incentives—anything you've seen in here about their potential for growth or expansion?
>>Pieter Gagnon: Yep. So, they are both here. Whenever they show up in the legend, it means they're there. It's just—right, it's small enough that you can't see it. And there are for both of these technologies things like CSP and geothermal definitely. I'm not actually sure about biopower in absence of CCS, but I know both CSP—concentrated solar power —and geothermal, we do actually see scenarios where you see some more of those technologies being built, especially out in California with their aggressive state policies. But it still ends up being so small that you can't really see it on a national level. I don't think that there is a scenario where we see whatever I would call my subjective opinion here of significant amounts of CSP or geothermal. We do see some scenarios where it does get built, but it's never a ton.
I will say that we are—a big geothermal project is starting up at NREL, so over the coming—I don't know when stuff will start coming out about that but over the next year or two we'll both be revisiting a lot of stuff about geothermal as well as putting out geothermal reports that will talk a lot about specifically what are the pathways for technologies like that? It's—let me just say, it's really hard to beat wind and solar and batteries with IRA incentives. They're really cheap.
>>Wesley Cole: Yeah, and I think that's a key piece, is a lot of technologies got incentives under IRA but so did a lot of other competitors. So, if everybody gets cheaper, then it may not shift the relative competitiveness.
Can you talk about the effects of IRA on demand growth and how that's represented in the scenarios?
>>Pieter Gagnon: Yeah. So, I have a couple of… sorry… moving slowly. We do—so, with IRA we did kind of a—oops—a relatively quick adjustment to our demand projections. Don't bother reading this. Basically, the green line was what we assumed national demand would be pre-IRA, and then we took an old study that we had done—this electrification futures study—and we derived a—what I've been calling light electrification, so it's lower than our moderate electrification scenario, and we put that in for our new assumption under IRA. I want to be clear: This is—that was just kind of a very quick thing we did where we took some previous work, we looked at IRA, and we said on balance we think this is going to have a net kind of light electrification effect.
There are a bunch of things in IRA for energy efficiency and stuff like that, so there definitely—don't take these projections too seriously. We did make our best professional judgment about this and nudged the numbers upwards, but there will be—there's already work ongoing to do a much more serious bottom-up estimate of the impacts of IRA on demand. So, there's both an increase plus the shape changes a little bit—the load shape. This is the load shape in 2020 in our model and then this is kind of the load shape in 2050 with IRA. So, you can see here that there's kind of higher non-summer loads from electrification and stuff but it's not dramatic.
>>Wesley Cole: Yeah. So, a question on the existing nuclear fleet. So, you noted that it—because of the PTC incentives it's not retiring in the near term but eventually those do phase out. And so, a question about are those plants then getting retired—do you see any retirements of plants after the PTC expires? Or, could you talk to the existing nuclear fleet post-IRA phase out.
>>Pieter Gagnon: Yeah. So, this is actually interesting. I would not have—my intuition would not have expected this, I guess. So, first of all, just to kind of highlight, is we do see—in a lot of our scenarios we see there's some retirements. Some of those are announced retirements, which then we enforce. But then, we see some economic retirements at some—but it's mostly staying relatively flat. There are a bunch of scenarios where we see maybe a third of the fleet going away or something like that for economic reasons in —after the existing nuclear PTC goes away. But there really is this phenomenon where it seems like at least our model sees the 2020s as being particularly difficult for nuclear. And if they can make it through that, then the value of capacity goes up enough that they're able to stick around and they're not subject to as many—as much retirement. So, if you didn't have IRA, there would be a lot more—if you didn't have IRA for nuclear, but only had IRA for other technologies, you'd see a lot more nuclear retirements. But we don't see that. And yeah, capacity values go up enough.
I will say one assumption that we do make is that we assume nuclear plants become—right now, almost all nuclear plants just operate full out all the time. It's technically possible to ramp them. So, we have it so that nuclear plants in the Cambium modeling, the stuff that we can do, can ramp down to 70% of their level. That helps a bit with integration of renewables, and they stick around a little bit longer because of that because you don't have to kind of have them either on or off. Anyway.
>>Wesley Cole: We have a few questions about natural gas. So, it's just questions about—so, both near term and we're showing relatively flat gas generally in terms of capacity, but there's still a lot of announced builds out there and a lot of activity going on. And then, separately, there's also—what does natural gas builds look like after the IRA—the tax credits phase out? Is it sensible to have those phase out and not—would other subsidies also phase out that could —that benefit fossils or that—it's kind of this relative competitiveness question. How do they track with each other? So, kind of two related questions there, that natural gas in the near term, especially with there's a lot of activity there, and then over the longer term with the phase out of the tax credits.
>>Pieter Gagnon: Yeah. So, I guess one thing just to emphasize is that even though we have the generation contribution of natural gas, we have kind of declining—there's this—in the 2030s there's kind of the low point for natural gas, the capacity doesn't decline. There's actually—we kind of lump in this lighter purple, it's both CTs—combustion turbines, the peaker plants, as well fuel oil generators. So, we actually see a little bit of decline there. I'm pretty sure that's almost all the fuel oil generators retiring. Some of those are pretty inefficient. But the capacity actually sticks around. I think that there might be a little bit of growth. There certainly is growth by the end. The overall capacity increases. We have some plots—you can download the data and look at that; we have some plots in our report that I didn't show here. So, it's—a lot of the story for natural gas is that it's not like we're building less of it. We slow down—we're not building tons more in the projections that we have, but we do use it less because of the deployment of renewables.
What I'll also say is, just to reemphasize, this is a least-cost optimization model to the extent that there's kind of inertia in decision making and resource planners continue to invest in fossil generators even in absence of—even if it's possible that renewables would be cheaper in their scenario. We're not—we can't —we can't—well, we could. Our model is not trying to anticipate that behavior. Yeah. So, this is a least-cost optimization model.
>>Wesley Cole: Thanks. Another question: Could you talk more about hydrogen? I think you mentioned it once or twice, but it's not clear—nothing on here is labeled hydrogen explicitly.
>>Pieter Gagnon: Yes. Yeah.
>>Wesley Cole: Can you talk more about hydrogen and how it fits in with everything here?
>>Pieter Gagnon: Yeah. I was waffling a bit on whether or not this year to change it from "renewable energy"—we call it either RE-CTs, renewable energy combustion turbines, renewable fuel combustion turbines. This is—in ReEDS we have what is just a generic plant that we are assuming has access to a renewably derived fuel, combustion fuel. That could be hydrogen or it could be biofuels. There's a handful of different potential technologies. I think—I am in no means an expert in the entire renewable fuel, the different potential pathways we have out there—it seems like it's probably—we have hydrogen plants already—pilot plants being announced and stuff like that, so those seem to be heading the game. But anyway, we just have a generic representation, assuming that you would be able to have a renewable-derived fuel for $20.00 in MMBtu, which is pretty high. That's certainly much higher than current-day natural gas prices, which is why you only see it showing up in small amounts here in some of these scenarios and mostly being used for peaking capacity, firm capacity. I will say that if you are—if you care about stuff like this, the role of these technologies, NREL does have other reports. We have a recent study that we put out, "100% clean electricity grid by 2035," something like that, I don't remember what the title is, that really dives into a lot more of what does a fully decarbonized grid look like and what is the role, potential role of these different kind of nascent technologies? And that has a lot more thorough treatment of these kind of renewable fuel technologies in it than in the Standard Scenarios. And so, if you care about that, I encourage you to look up that 100% clean electricity study that NREL put out recently.
And Wes, feel free to add anything if you have any additional thoughts at any point.
>>Wesley Cole: Okay. Can you talk about long-duration storage related to hydrogen here? What's—people are asking about—does the model deploy long-duration storage? Is that represented?
>>Pieter Gagnon: Yes, we do not have long-duration storage represent in the model, other than you could interpret the RE-CTs as quasi-long-duration storage because you would potentially be making those with electrolysis at certain times of the year and using it at other times of the year. I want to be clear: In Standard Scenarios we do not actually add the load for the hypothetical electrolysis that would be creating the RE-CTs. We felt okay with that for these scenarios because the amount was so small that it would not—it just—we didn't—the tradeoff between including it and not including it, we tipped towards not including it. However, in other scenarios, like that clean electricity study, I'm pretty sure that they have the actual load from the electrolysis represented in the modeling, which therefore is a lot closer conceptually to a—you're storing energy essentially in the fuel. But we don't have long-term battery storage. We have work going on in that area but it's not in the model yet.
>>Wesley Cole: With the caveat that it depends on what you define as long term. Up to 12 hours is in the model.
>>Pieter Gagnon: Yes. Yeah, thanks.
>>Wesley Cole: So, if you think long term is less than that, then it's there. Can you talk about—I know you've talked about CO2 emissions. Are those CO2 only? Do they include upstream emissions? Other—can you talk about what that means in terms of CO2 equivalent or not when you're doing carbon reduction?
>>Pieter Gagnon: Yeah. So, when I—when we were plotting the—this, I was just showing—this is just only CO2 emissions from the direct combustion of fuel. So, it's not CO2e. It doesn't have precombustion, methane leakage, and stuff like that. And our decarbonization trajectories are defined in terms of just CO2. They're not—so, when we say net zero it's not taking into account methane leakage or upstream fuel—I should be broader—it's not taking into account the emissions associated with upstream activities like construction or fuel extraction processing and transportation. We do actually now – new metrics that we are reporting is that we are reporting those emissions, both precombustion—it's only—right now, we're only reporting the precombustion, the three major greenhouse gases—CO2, nitrous oxide, and methane—for fuel extraction processing and transportation. We have that in the data set. I didn't show it here. And we don't define our trajectories based off of those numbers, but they are there if you want to—so, you can plot CO2e including upstream emissions if you want. Just go and download the data. And then, we also have SOx and NOx this year. We did some improvements and got to the point where we were comfortable releasing SOx and NOx, which are not greenhouse gases, but emissions.
>>Wesley Cole: Another question about the effect of the IRA on technology progress—so, we're seeing a lot more wind and solar. How does that impact—hypothetically there will be some learning in there. You get cost reduction as you employ more technologies. Can you talk about how that gets represented or not in this work?
>>Pieter Gagnon: Yeah. Right. So, things like technology learning but then also other effects, like we have some pretty rapid buildouts. None of them are too wild. Well, I guess, some of the CCS, there's—a lot of it is suddenly appearing. But wind and solar, there's aggressive growth. It doesn't—my subjective feeling is that it's not unreasonable growth rates. It's definitely aggressive. Anyway. We have high rates of growth for different technologies and we don't explicitly represent supply chain effects or anything like that, but we also don't represent learning from doing a lot of deployment of wind and solar and other technologies and storage and everything else. All we do is we take—for technology costs, NREL has annual technology baseline numbers that they update each year. A lot of effort goes into those projections and we take those external projections and then put them in the model and see what gets built.
So, there's a lot of stuff there that we—that could be impactful. The technology learning, there's things that could both push those up as well as push them down, the costs. And we don't model that, which I used to feel better about. Now with some of these growth rates, I wish that we had—that we were—at least had some scenarios where we were looking at the potential effects of technology learning and stuff.
I will say that our—the nascent technologies, things like the bioenergy with CCS, small modular reactors, all this kind of stuff have declines over time that we assume, which is kind of an implicit assumption of things like pilot plants happening in the U.S. and technology deployment in other parts of the world, not in the U.S. So, even if a technology is not being deployed in our scenarios, it is still—we are making an assumption that there are cost declines over time based on these external, exogenous analysis efforts.
>>Wesley Cole: And do you want to dig in any more detail about the cost assumptions? That annual technology baseline is the basis where these come from and they provide a lot of information about how they develop costs for those, so I would recommend that if you're interested in the cost side.
>>Pieter Gagnon: Yeah. The cost as well as some performance assumptions and stuff like that. It has things like—other things like that. It's a great resource, the NREL Annual Technology Baseline we've been putting out for—I think this is also the eighth year for it. So, yeah, a really good resource.
>>Wesley Cole: So, we have a few questions related to Cambium. So, one is about the short-run marginal emission rate versus the long-run marginal emission rate, and the question is how much curtailment would a region need for the short run to be more accurate than the long run? Or, maybe more generally, how do you—are there ways you could think about which one you should be applying? And I know you could probably talk about that for an hour, but if you could give a one-minute summary?
>>Pieter Gagnon: Yeah. I'm going to—because we don't have too much time and I want to make sure that we can get to other things, I'm not just going to espouse about emission rates. I will say that we've written about them, so if you really care about them, especially marginal emission rates, short-run versus long-run, look both towards that paper that we put out—we have workbooks where you can kind of plug in some assumptions about the time frame that you care about and get the marginal emission factors spit out for you. It's a little bit easier to work with than the CSVs.
I'll just generally say that for—if you're trying to comprehensively describe the impact of an intervention, then long-run is usually the way to go. If you're looking over a 20-year period, there are very few scenarios where a short-run would be the appropriate way to describe our best guess of the total impact of that technology over that time period. If anything, there are scenarios where if you're not going to build anything more, then the two numbers just converge. But we don't really see that happening in the U.S.
So, overall, if you're trying to comprehensively describe the impact of an intervention over a long period of time, long-run is almost always the way to go. However, short-run is both—(a) it's appropriate to use short-run to describe kind of the immediate impacts of a technology, especially if it was not anticipated by resource planners, so sometimes blending the two methods is the best way to go. There's also a role for the short-run perspective in kind of real-time decision making. Like, the idea of "When should I charge my electric vehicle?" would ideally be a combination of both what's happening right now, as well as how does this decision influence the ongoing evolution of the grid, some type of combined metric that currently doesn't exist. So, anyway. I'm getting ahead of myself.
I did see a question; I'm just going to hop in and say someone asked about the timing for the Cambium data sets. As soon as possible. We try to get them out in late November, early December, so we're behind schedule because of IRA. It put everything back. It's either going to be late this week or sometime next week that the Cambium scenarios will be coming out. So, feel free to either email me or just check the websites, or there should be an email blast going out about that.
>>Wesley Cole: Can you talk about the resolution of those Cambium data sets? I think you mentioned it briefly, but just the difference between the Standard Scenarios and Cambium in terms of their resolution?
>>Pieter Gagnon: Yeah. So, the Cambium data, a lot of the metrics are available at hourly resolution through 2050. It's not every year. It's only years within that time frame, but hourly data. And then, the geographic resolution is you can download it at 134 different—we call them balance scenarios but that's —it's not the same as real-world balance scenarios—134 different nodes in the US. In most instances, you don't need to go down to that resolution, especially the emission metrics. You can work with larger regions and you can download the data at larger regions. So, yeah. I think that's probably what they were looking for there.
>>Wesley Cole: Okay. And probably time for one more question. So, can you talk about costs in these scenarios and what's included in there?
>>Pieter Gagnon: Sorry, costs?
>>Wesley Cole: Are there cost metrics included in the viewer—
>>Pieter Gagnon: Oh, yeah. Yeah.
>>Wesley Cole: —that people can look at?
>>Pieter Gagnon: Standard Scenarios does not have cost metrics, either average or marginal costs. Technically, if you're an academic and you have the time to dig in, you can download the data, the repository that we use and you can actually either rerun the scenarios and then you get all of the data or you can look at the assumptions we had and kind of plug in our numbers. You can go and look at the costs that we assumed and multiply it by the capacity that gets built and kind of reconstruct things. Some people have done that. It takes a lot of work. I'm not saying it's a good idea. So, my apologies. We should—I wish that we could be packaging more cost metrics into the Standard Scenarios. It's not there. There are a handful of cost metrics in the Cambium data sets, mostly marginal costs. Still not average costs, which is just—we only have so much time and we haven't packaged that stuff up yet.
>>Wesley Cole: I do remember a few cost plots in the report itself too, so I think there is at least some high-level…
>>Pieter Gagnon: That's true. We do plot in the report. There is a marginal cost of energy and capacity, which is actually not available in the electricity—the stuff you can download, which is just—it just wasn't packaged into the CSVs. But it is in the report.
>>Wesley Cole: Okay. Well, we're at time. I think either Pieter or I would welcome if you have follow-up questions. You're welcome to send us a note and we'd be happy to follow up and chat later if you wanted to. But we do want to make sure we're ending here on time. So, anything else you want to add, Pieter, before we close?
>>Pieter Gagnon: No. Thanks, everyone, for joining. And yeah, don't hesitate to reach out if you have any question about the scenarios. But check the report first. So, all right. Thanks, everyone.
[End of Audio]