Beyond Technical Potential: The Challenges of Siting Wind in a Low Carbon Future
Alex: All right, well, I think we can go ahead and get started while everybody's kind of settling in. I wanted to welcome everybody to today's webinar. My name is Alex. I'm really honored to be a part of this Wind Energy Science Leadership Series.
It's been going on for quite some time now. It's an ongoing series of educational webinars that include presentations and discussions on many wind energy-related topics, featuring speakers from our laboratory, strategic partners, and other thought leaders from the energy industry. These webinars discuss the challenges facing wind energy and the pathways forward for making wind, one of the most prevalent energy sources of the future. In today's webinar, NREL analysts, Eric Lantz, Trieu Mai, and Anthony Lopez will present two related research efforts that use detailed geospatial and power sector modeling to shed light on the interaction between siting considerations and clean energy development.
But before we begin today's webinar, I wanted to let everybody know it will be recorded and available on NREL's website. Please make sure you mute your lines if you're not speaking today. We encourage everyone to use the raised hand feature or just enter your questions into the chat, and we'll be answering those at the end of the session. Eric Lantz will be moderating that. Finally, I would like to introduce you to Eric Lantz, a co-author of this research to give a quick introduction of what you can expect to hear today.
Eric? Eric: Great. Thank you, Alex, and thanks to all of you participants for joining our webinar today. As Alex mentioned, I'm Eric Lantz. I'm the NREL Wind Analysis Platform Lead, and I'll be moderating today's session. Just very briefly, on the logistics front, we're going to try and handle questions today through the chat feature within Teams.
Hopefully, that's accessible to everyone. It looks like it's working at least for a few of our external participants. So, hopefully we can handle things through the chat there. If needed, we can also open the mic to verbal questions from the group if you're unable to access the chat later during the Q&A session. As Alex indicated, we are recording this session, and we'll post a recording of the webinar along with a transcript on the Wind Energy Leadership Webinar Series website within a week or two, most likely, of completion today.
With that, our topic today explores the relationship between our abundant wind resource and the challenges that might come as we seek to leverage this tremendous resource to serve an energy future where 40 percent or more of our electricity needs and, effectively, a very large share of our overall energy needs are served by wind power. When we think about the technical potential of wind, we often hear numbers that are a factor of five X or more than our current U.S. operating installed electrical generation capacity. We have a similar magnitude of available wind energy. In a sense, the wind energy that's available to us far exceeds what we think we might need in terms of electricity or energy. But in reality, there's a non-trivial risk that this affective available resource, what we can actually build and deploy commercially may be substantially less than our technical potential estimates.
We hear about this from the developer community. Often they're frequently frustrated by an array of market, transmission, and siting challenges that preclude them from accessing prime wind energy areas. In addition, issues like social acceptance and wildlife, which have been part of the wind energy story for decades, are really compounded when you have increased deployment and the potential for cumulative effects.
Over time, and especially in the past, decision makers and researchers, like us, haven't been too concerned about these risks, mainly because what were perceived as the barriers to wind energy deployment were primarily economic or, in some cases, the power systems consideration. So, the grid integration considerations. But today, we find ourselves in a very different place.
The cost of wind energy and other renewables is falling rapidly. For wind specifically, we have a great deal of resources that's available within the two- to four-cent per kilowatt hour range, even when you take away the PTC. This low-cost wind is available in broad geographic areas, many of which have historically been considered non-windy areas unlikely to see commercial development. So, the landscape is really changing.
A lot of these regions are becoming more accessible to commercial wind activity. Moreover, the urgency of sort of decarbonizing our energy system and the economy has become more critical as populations in the U.S. and around the world are starting to experiencing climate change in relatively palpable ways. So, within the research world, we find ourselves exploring scenarios that ten years ago weren't even on the radar screen and, five years ago, were still pretty implausible over the time frames that we expected to see deployment occur over the next couple, three decades out to 2050. In addition, we're getting just more and more better insights about how much wind might be needed in high electrification and deep decarbonization scenarios. So, as we kind of wrestled with these different factors at play, we decided we could no longer use basic or simple technical potential estimates to inform the research that we're doing about how wind energy might serve the future of the electric power system.
So, we embarked upon this path to kind of create greater resolution and greater detail, greater granularity into our technical potential estimates. Then we've started to put that through the paces of our capacity expansion analyses. Today, you're going to get to hear from our two leaders in this space, Anthony Lopez and Trieu Mai. Maybe, Anthony, you want to just go to the next slide? Yeah, great. Both Trieu and Anthony have been doing this kind of work for over a decade.
They're both international leaders in this space, having done work for the International Energy Agency and the International Renewable Energy Agency. Two leaders at the global level here. Then, their work has also been used for analyses and related questions up to the highest levels of the federal government. So, Anthony's going to first walk you through our new and latest technical potential estimates. Then, Trieu is going to take us through our capacity expansion results. Before I pass the baton to Anthony, go ahead and take us to the next slide.
We're going to talk just really specifically about the research questions that they're going to focus on. The first is really, "What is the technical potential of U.S. wind power and how sensitive is that to different levels of siting restrictions and technology evolution?" We actually pose this question, "What is the technical potential of wind power in the U.S.?"
I actually think of this as a little bit of a misnomer because I think, actually, technical potential is something that's dynamic and subject to change as technology evolves and as society grapples with utilization of the resource in different ways. But Anthony will illuminate that for you very clearly. Then, Trieu is going to take us through those longer-term capacity expansion results, to think about, "Okay, given these different considerations, how might that affect the location and the quantities of wind power that are deployed across the country?" So, without further ado, let me pass it off to Anthony here, and he can take you through our first portion. Do feel free, please, to put your questions into the chat. I'll be keeping running tabs on those, and we'll get to them later on in the Q&A.
Go ahead, Anthony. Anthony: All right. Awesome.
Thanks for that introduction, Eric. So, I'm going to be walking you all through the geospatial supply curve modeling that explores various siting constraints and turbine technologies and their influences on wind potential. Really, to get at the question of just how much wind potential does the United States have. Just a note here, you can see even more details of this work in our journal article published in Energy. Okay, so first, I want to introduce you to the Renewable Energy Potential Model, or reV.
It's an opensource model developed by the National Renewable Energy Laboratory. We use reV for this analysis, which is a multi-criteria decision support model that integrates high-resolution spatial temporal wind and solar resource data, amounting to tens of terabytes of data. This is combined with the Systems Advisor Model, which estimates the generation potential of a given resource or a given turbine and the site-levelized cost of energy. For folks on the line, can you please mute? reV also has a spatial module used for determining site accessibility defined as technical exclusions.
Further, reV estimates the levelized cost of transmission to evacuate the site dependent energy to the electric grid. Finally, reV has a module to package the data, including the power profiles associated megawatts of capacity and the spur transmission cost for use in a variety capacity expansion models. But for this study, we're piping the results to reEDS, and that Trieu will cover in a little bit. So, this slide represents the step change in our ability to capture spatial constraints that wind developers might face when developing a project. Historically, we've relied upon course land-use/land-covered data to represent the built environment or areas that would preclude the placement of a turbine, for example, on a road or in dense urban environments. However, as you can see in the image on the left, the land-use/land-covered data only capture the densest developments, which is limited by the minimum mapping unit associated with this data.
We also did not historically capture potential setback requirements likely adhered to by many developers to limit placement of turbines next to homes, too close to electric transmission line, or roads. So, the map on the right demonstrates our enhanced spatial resolution through the use of new vector mapping of the built environment, including every building footprint in the United States from the Microsoft Buildings Dataset and every road, transmission line with their estimated right of ways, railroads in the United States from a variety of sources. You can see by the plotting of the existing turbines from this wind farm, they closely match the spatial constraints adhered to by this wind developer. Now, to do this national assessment, we had to partition the country using a fishnet. We chose an 11-and-a-half-kilometer grid, which represents the typical area of a wind plant extent.
This results in roughly 67,000 individual wind sites across the conterminous United States. Each one of these sites is characterized by the distance to transmission generation potential, capacity, LCOE and more. Okay, the previously slides, they demonstrated the spatial fidelity of the modeling that we're using. So, now, we'll move into this particular study, its design, and the results.
So, first, we designed scenarios around three turbine cases and three siting regimes. I'll discuss the siting regimes in a couple more slides. The three turbine cases captured today's technology in the current case, while we also explored turbine designs that may come about in the next decade through the annual technology baselines moderate and advanced turbine designs.
You can see in the scenario matrix that we modeled the reference siting regime for all turbine cases and modeled the three siting regimes for the moderate turbine case. Okay, first, I'm going to go through and define the philosophy of each of these siting regimes, and then we'll explore different features of the map. So, open access is the least restrictive scenario, applying only physical barriers and protected lands – for example, national parks and conservation easements – that restrict wind development, setting a ceiling to the available land. Note that the open access siting restrictions serve as the basis for all other scenarios, meaning we add exclusions to them.
The map shows the available capacity when accounting for siting criteria. So, the darker the blue, the more capacity available at each one of those 11-and-a-half-kilometer sites. So, some interesting things on the map here. You can see in Colorado, mountainous landforms and high elevations excluded. So, areas above 9,000 feet. We can also see water and wetlands being excluded or reducing the capacity potential in Florida, the Coastal Carolinas, and Georgia.
We can also see the protected areas showing up in Yellowstone and up in Wyoming. We can see we exclude urbanized areas, airports, and existing infrastructure as it apparent in the Chicago area. Okay, moving onto the reference access.
Reference access builds upon spatial exclusions defined in the open access siting regime, and is a moderate representation of exclusions that utilizes, where feasible, best management practices to guide development. So, to help guide these, we developed a database of all wind siting ordinances, which include 271 documented state and county ordinances pertaining to height restrictions and setbacks from buildings, railroads, waterways, and transmission lines. We apply these individual ordinances for their jurisdictions, but also apply a setback requirement of one-point-one times the tip height of the turbine for locations without ordinances. Note that a setback requirement of one-point-one represented the majority within the siting ordinance database. I would also like to mention that we captured even more than just height and setback requirements but did not model them.
These include ordinance like noise, flicker, and land property setbacks, and are freely available for download at the link below. So, to highlight some interesting things about this map here, probably the first thing that's going to pull your eye is the exclusions in Southern California. So, these come from documented ordinances for height restrictions in some of these counties that were below our turbine heights that we modeled. Next, you can see the civil infrastructure setbacks. So, the one-point-one times the tip height for roads, buildings, transmission lines, and railroads really eliminates a lot of the capacity – reduces a lot of the capacity I should say – reduces a lot of the capacity in the Eastern United States. This is really due to the kind of dense infrastructure and population density that we have comparative to the Western United States.
We also exclude radar proximity. So, four-kilometer for NEXRAD radar and nine-kilometer for short-range and long-range radar as established in previous literature. Then, finally, we exclude slope threshold.
We exclude all lands that have a slope greater than 25 percent. Do you see that's inverted? But we exclude all lands that are greater than 25 percent as established by the Bureau of Land Management for guidelines for erosion concerns. Okay, limited access.
It's a combination of the most – this is our final siting regime, and it's a combination of the most stringent siting considerations for all scenarios that represents a plausible floor for developable area. There are a couple of key exclusions applied in this scenario. First, we increase the setback requirement for civil infrastructure from one-point-one to three X the tip height. This amounts to about a 620-meter setback. Now, this – while this does seem stringent, there are 20 counties in the United States with a three X or greater setback for wind turbines.
In addition, there are even greater setback requirements in Europe. We also exclude radar line-of-sight view sheds or view sheds. While turbines can technically be sited within radar view sheds, it is plausible to saturate that view shed, leading to a restriction of new wind power plants in that area. We really wanted to explore this as a possibility moving into the future.
Finally, we exclude all federal lands. Now, while federal lands are technically available for development, there has been very little development taking place. We really wanted to capture this reality in our models and explore the implications.
So, now, some of these exclusions, they spatially overlap, meaning that the total excluded land for each category is less than the aggregate. We see in the open access regime that wetlands represent the largest exclusion. We see in the bar below the aggregate is less than the total exclusions. Again, this is just showing that spatial overlap. This becomes more apparent in the reference siting regime where roads and buildings setbacks represent the largest exclusions, but the aggregate is almost half of the total.
Finally, in the limited access siting regime, we see that three times the setback for buildings and roads represents the largest exclusions individually, followed by the federal lands and wetlands with the aggregate less than half the total. Okay, so what does this all mean in terms of total wind potential? Well, for the open access, we found the U.S. has about 15 terawatts worth of potential, while the reference access, again, following best management practices where feasible, has a potential of 7.8 terawatts. Under strict siting restrictions under limited access regime, we find about two-point-two terawatts worth of potential. Interestingly, the shape of these resource capacity remains largely the same in terms of wind speed.
But this is likely to change if we were to start including other siting considerations, including wildlife constraints that are more likely to be happening in the Western United States. In terms of all in LCOE or a combination of the site LCOE and the levelized cost of transmission, we see the shapes remain the same across siting regimes, but that when looking at the total available capacity below $30.00 per megawatt hour LCOE, we find 2,000 gigawatts in the open, roughly 1,300 gigawatts in the reference, and 433 gigawatts in the limited access regime.
Note that this is the for the moderate turbine case. So, these do represent cost estimates in the year 2030. So, in reV, the levelized cost of transmission, or LCOT, considers the distance in dollars per megawatt kilometer, and the point of interconnection in dollars per megawatt for all available wind sites. Site-based LCOE and LCOT are computed independently and added together to represent "All-in LCOE".
Note that we do not evaluate bulk-transmission upgrades as this is handled in reEDS. So, in the graph at the top, the solid line represent the All-in LCOE, while the dashed lines represent the site LCOE, and the delta represents the LCOT. So, we can see it can have a pretty large influence. For the reference access siting regime, we found a mean LCOT of roughly $3.00 per megawatt hour, which is in line with empirical data recently reported by Gorman, Mills, and Wiser.
All right. We also found, unsurprisingly, and definitely influenced by our wind siting stance, that siting regimes influence wind plant sizes. You can see those on the distributions on this bar graph. We also found that their regional trends are also apparent in all siting regimes with the Southcentral and Southwest having the largest plants compared to the smaller plant sizes in the Northeast and Southeast. Again, this can be easily explained by the increased population density and this increased civil infrastructure in the Eastern United States.
Okay, moving on into the turbine technology influences on the wind supply curve. We find decreasing LCOEs from the technology advancements across all regions of the United States. This is driven by both capacity factor improvements seen on the left and lower turbine costs which shift resource to lower LCOEs on the graph on the right. In terms of regional cost impacts for changing technology, the relatively static nature of transmission costs, at least in our modeling, results in there becoming an increased share of the overall cost of wind as turbine technology advances and turbine costs are reduced. We also find that less transmission dense portions of the Western United States have proportionally larger transmission costs, which is a result of when resources requiring longer-spur transmission lines to tap into the existing grid infrastructure and, in some cases, with very low site-specific wind plant LCOEs. Okay, so taller turbines, they require greater setback distances because we're using this tip-height multiplier.
This has an effect of reducing overall capacity 14 percent and 20 percent for the moderate and advanced technology as compared to the current technology respectively. We can see this in the bar graph where the orange bar represents currently technology as a – where the orange represents current technology and has a higher capacity than the green moderate or blue advanced turbines. However, generation potential remains roughly the same nationally, within one percent. This is really due to increased hub heights, tapping into better resources, getting higher capacity factors, and lower losses as really the specific power of these different turbines is not significantly different. We can also see this in the black dots on the graph.
We also find some interesting regional trends in these results. So, where in the Great Plains, the amount of capacity potential is significantly more than that of the moderate and advanced case, but the generation is more for the moderate case. In the mountain, again, the current case has more capacity but also has the highest generation potential. Now, this is likely due to the fact that we deploy a single turbine design across the nation or, rather, we do not optimize the turbine design spatially for different wind resource regimes. Our intention for this body of work was really to tease out these differences through scenario basis and to try to find these kinds of differences. But we do have plans to explore spatially-optimized designs and future work.
All right, in summary, the U.S. is rich in wind resource with a ceiling of 15 terawatts of onshore technical potential, but siting restrictions could substantially reduce land accessibility, resulting in roughly 2 terawatts of technical potential in our most constrained case. Sizeable setbacks for building and other infrastructure could limit wind resource availability, particularly in high-demand regions, and they require spatially-explicit modeling to tackle.
Wind turbine design can also influence the capacity resource potential through interactions with setbacks, but these need to be weighed with cost and generation potential. So, in closing, the video on the right shows our interactive web map that allows anyone to explore the data. This can be accessed at the link below with the video, as are the supply curve data, which are available for download as well.
So, I want to thank you all for your time today. Now, I'll hand it over to my colleague, Trieu Mai, who uses these siting scenarios to explore implications for the evolution of the U.S. power system. Thank you. You can go ahead and take over. Trieu: Thanks, Anthony.
Let me move this over here. Looks like there's a hand up. Carl, do you have a quick clarifying question? We'll have time for a broader Q&A later.
Audience: Yeah, I can save it for later. I didn't realize there was a second half to the presentation. So, please go ahead. Trieu: Okay, thank you. Anthony, you can see my screen right now, full screen? Audience: Yes, sir.
Trieu: Great. Well, thanks, Anthony, for that presentation. So, for the second half here, we're going to explore some of these power system scenarios. I've seen some of the exchanges on the chat.
So, hopefully, this will answer some of those questions. But before doing so, I just want to take a step back to what Eric said earlier. We've looked at this issue before. We've looked at in kind of course heuristic forms. We've always thought, or at least I've always thought before embarking on this that we know that the wind potential is large. You've seen some of the terawatt-scale numbers that Anthony presented.
We've always thought that, yes, we may be off in terms of the unexpected resource potential, but in large faction of a very large number is still a very large number. So, perhaps it doesn't matter. The way we've applied it in the past, it hadn't really mattered to our scenarios.
However, the combination of the work that Anthony has done looking at some of these restrictions and land-use conflicts, potentially, at greater detail have both realized that maybe our very large number wasn't so large or isn't so large. You can see that in some of the limited access results that Anthony presented earlier. On the other side, the demand side, perhaps the demand for wind energy in the future will be much greater than we anticipated, even five, ten years ago, as Eric alluded to, and in the context of looking at the conversion series. So, that's what this work is about.
This is using what Anthony presented, applying it into our power sector models that valuate the demand for wind and other resources, and for this specific analysis, looking at the siting restriction question is, "How much does it really matter? Does it really matter?" With that, I'm going to advance here. To do that, we use our flagship capacity expansion model for the power sector, reEDS or the Regional Energy Deployment System model. I won't go into any sort of detail in terms of the model methodology. Happy to answer questions at the end or feel free to reach out to me.
The key component here is reEDS identifies the optimal portfolio across generation, transmission, storage, mixes from today through 2050. It does so, and how we are uniquely using it for this analysis is because it has very high spatial resolution for its class of models. You can see a map of that on the left-hand side.
On top of that, it considers the various aspects of different generation technologies. The production profiles, the transmission needs, and the like. So, what we're doing is using this model to look at the impact of siting – so comparing those different siting regimes that Anthony presented – on the magnitude and location of future wind deployment, electricity prices and other electric system costs, on power sector emissions, as well as the cost to abate CO2.
So, to examine that, again, we're really – the core thing that we're analyzing are differences between those three siting regimes; open access, reference access, and limited access. They differ in technical potential by significant measures. Open access was roughly 15 terawatts. Reference access was roughly eight terawatts.
Finally, limited access was two terawatts. So, still, we see a significantly larger demand for wind today, but perhaps, in the future and some of the locational aspects can be important. But those differences between those three siting regimes could vary across what future energy system one might be interested in. So, to do that, we evaluated three buckets of scenarios.
The first one is a business-as-usual – today's policies only. Within that bucket, the BAU scenarios, we did look at two different variations in turbine projections, turbine cost projections. Eric: Trieu, somehow you muted yourself. Audience: You're on mute. Trieu: Looks like somebody muted me.
But back on here. So, yeah. Within the BAU group, we did look at two separate turbine layouts – ATB Advanced and ATB Moderate cases. In the second bucket, we looked at a set of low emission scenarios where we applied a power sector emissions CO2 cap at 80 percent by 2050. That's 80 percent relative to 2005 levels.
We ratchet that down to a 95 percent emissions reduction level by 2050 in 5 percent increments. The reason why we're doing this is we're systematically increasing the stringency of the carbon cap to see where the impacts of wind siting might change kind of qualitatively. Lastly, we looked at a set of scenarios where we prescribe the amount of onshore wind to 40 percent by 2050. We'll get to the justification and motivation for those set of scenarios as well as other sensitivities associated with them later. So, let's get started on first BAU scenarios to get us grounded. On the top chart here, you should see the reference access BAU serials for moderate technology assumptions for wind on the left-hand side, and advanced technologies.
Sorry, the X axis is not shown, but that's 2020 through 2050. We won't really go through the nitty-gritty generation mix on these scenarios, but qualitatively, both of them show a decline in fossil generation over time, a decline in nuclear generation as a result of plant retirements, and an increase in renewable energy displacing some of those generation decline as well as meeting new demands. Most of the renewable energy growth is from solar and wind. The relative mix between those two depends on your cost assumption. So, that's consistent with many other analyses that we have done, and others have done. What's interesting, of course, is not so much the sensitivity to technology cost assumptions in this analysis but, instead, the impact of the different siting restrictions.
So, now, in this second row, you see the difference from the records access for the corresponding scenarios using the open access where the resource potential for wind is effectively double compared to the top row there. Not surprisingly you see is an increase in wind deployment as a result of that. There are higher-quality wind sites that are available under open access that weren't under reference access and, hence, you get a boost in wind generation. This increase is roughly four or five percentage points by the 2040s. So, by 2050, you're looking at a serial that's almost 40 percent onshore wind generation in the U.S. power system. The tradeoff between wind is mostly with fossil generation.
You can see that in the negative side of the bars on the right, as well as a little bit of solar generation as well. Now, the same comparison is made with the limited access scenarios and, obviously, the signs flip here with more restrictive siting for onshore wind relative to the reference access. You can see a bigger – a difference and even a bigger in magnitude difference such that with limited access, you're seeing a reduction in wind generation by about 10 percentage points by 2050. So, really, a significant impact well beyond what our prior analysis have shown.
These are just the BAU scenarios. So, what does that correspond to in terms of capacity? So, focusing on the darker shades at the top of these charts here are the advanced technology cases. You can see with reference and open access siting, about 500 gigawatts of onshore wind by 2050. So, kind of steady growth and pretty sizeable growth over time, a little bit accelerated from what we've seen in the past ten years or so through, again, 2050.
In comparison, however, the red line with the limited access as you see a significantly lower deployment level. In this scenario, as we saw in the prior slide, it's roughly a 25 percent wind generation and roughly 330 gigawatts. So, a gap of about 100 to 200 gigawatts between the limited access case and the other cases.
Again, everything else is the same. Wind technology assumptions are the same. Policy conditions are the same. It's simply in effect of the siting restrictions. Now, wind, obviously, extends not to just onshore but also to offshore. I saw some questions earlier about are we treated offshore.
We haven't done the same analysis in terms of looking at the siting aspect with offshore, but we do have offshore wind in our model. What you notice here is, by restricting onshore wind, we actually see a corresponding increase in offshore wind. Sorry, corresponding is probably not the right word here. An increase in offshore wind that, in part, makes up for some of that reduction in offshore. Be careful about the scales here.
This is an order of magnitude lower Y axis scale than the onshore one. So, you certainly don't make up all of the difference there. But we do see this balance between onshore or this tradeoff between onshore and offshore wind deployment, which we'll discuss in a little bit. Beyond wind deployment results, we also look at electricity price and CO2. It impacts us well. So, electricity prices here on the left.
These are wholesale prices. CO2 emissions from the power sector are shown on the right. Again, unsurprisingly, you expect higher prices with more restrictive siting because some of those low-cost resources are unavailable and, similarly, with less wind deployment, you'd expect higher emissions. That's both borne out here.
What's interesting to me is the magnitude of those impacts. Focusing on the electricity price results around the 2050 time period, you can see sizeable $5.00 per megawatt hour, up to $5.00 per megawatt hour differences across the siting regimes, which is similar in magnitude to the impact of comparing a moderate technology cases with the advanced technology cases.
In other words, siting can be just as important as driving down costs through R&D for wind in these scenarios, at least. As well on the CO2 cases, you can see that the impacts are sizeable, over 200 million metric tons by 2050. So, really significant CO2 impacts. These are all BAU scenarios without an underlying national carbon policy. Okay, so that gives us a transition over to our low emission scenarios.
So, we're focusing only on 2050 for these results. Capacity on the top, generation on the bottom. The bars show onshore and offshore wind – onshore in the lighter color – for the BAUs that we just discussed as well as the increasingly stringent carbon cap scenarios; 80 percent to 95 percent. No surprise here.
Reference access, no surprise. Here, with increasing stringency on that cap, you get more wind. Wind is one of the leading clean energy technologies and the costs are lower than many others. Hence, you get an increase where in the 95 percent CO2 scenario over 700 gigawatts of onshore wind are deployed, making up about 45 percent total generation. But what's interesting in our analysis is how that compares with the different siting regimes.
In the middle column, shows the open access. The differences are modest, but directionally, you do increase the amount of wind relative to the reference access. In the 95 percent scenario for open access, we're looking at 800 gigawatts of wind combined onshore and offshore. What's most interesting is this limited access case here. We see qualitatively the same trend. Tighter carbon cap, more wind.
However, you start to see a saturation of how much wind you get, especially looking in generation terms. You start to see that even as you increase the carbon reductions to 90 percent, 95 percent, you're still having a hard time reaching 30 percent onshore wind. In fact, even with the 95 percent emissions reduction scenario here, the amount of onshore wind with limited access is less than in the BAU cases for both reference access and open access. So, really strong signals and sensitivities on onshore wind deployment coming from the limited access scenario. However, we're going back to the offshore story, earlier, again. So, the darker blue bar, on the top of the bars here, with limited access and with that carbon constraint on there, you can start to see a really significant deployment offshore such that for the 90 and 95 percent scenarios, we're looking at over 100 gigawatts of offshore by 2050 and over 10 percent generation shares in the most stringent cases.
So, the U.S. is rich in clean energy resources, whether or not they're onshore or they're offshore, they're wind or solar. But there are cost implications associated with that, with using a different generation mix as well. So, one measure that we look at on that cost side is the marginal abatement cost. I'm only going to focus on the top row for this presentation.
Marginal abatement cost is the cost to abate the next ton of CO2. First, we start with the reference case. You can see these abatement costs increasing both over time and it's a function of the stringency of the carbon cap. Neither of those effects are surprising. You do see this non-linearly ramping abatement cost as the carbon cap get tighter and tighter. Again, are you're approaching zero carbon, it's going to be harder and harder to avoid the next ton.
The low-hanging fruits have all been picked and therefore, it's going to be more challenging. However, for the reference access case, you can see that the abatement cost, even at 95 percent CO2 emissions is below $60 per ton in that 2050 year. With open access case, the abatement costs are very similar, within $5.00 per ton of CO2. In most of the cases, a difference of within $5.00 a ton compared to the reference access case.
So, now big surprise with the other results. Again, what's interesting is the limited access case where those constraints on the siting could really raise the challenge, the cost to abate further emissions. For the 90 and 95 percent cases, we're looking at $20.00 per ton higher abatement costs compared to the other two. So, significant challenges there.
One side note I'd like to make, though, is if you look at these charts, what pops out to me is both the magnitude of the result for the earlier years – so prior to 2040 – abatement in the power sector is relatively cost effective, even with constrained onshore wind siting. In addition, if you look at the 80, 85 percent scenarios, abatement costs remain below $30.00 per ton, again, even with constraints, perhaps relatively severe constraints on onshore wind siting. Okay, so we've focused a lot so far on the national results for the BAU and low emissions scenarios. What we haven't looked at is the geographic distribution of wind deployment.
To do that, we wanted to isolate that geographic question by normalizing the amount of wind across these different siting regimes. Hence, we looked at these 40 percent onshore wind scenarios by 2050. What you see here is the reference access case. So, in that case, you're looking at 550 gigawatts in order to reach that 40 percent level. Forty percent is roughly the level of wind generation that we observe in many of the prior low emissions scenarios as well as many other studies have focused on.
So, what is the geographic pattern of wind deployment look like under such scenario? So, that's show on these two maps. 2020 is shown on the left for comparison. It's not quite the end of 2020, but it's close to it. So, 160 gigawatts of wind deployment mapped onto those 11-and-a-half-kilometer reV sites that Anthony mentioned earlier. On the right is this 40 percent scenario for reference access in 2050.
So, that's a five X increase compared to the 2020 levels. So, you can see one of the things that stand out here is to get to 40 percent compared to today requires geographic expansion of wind development extending beyond the wind belt into the Midwest, into the coastal states, into the Appalachian states, and even into the Southeast. So, really, with records access, you see this spread in wind deployment across the country to get to 40 percent. So, now, when we compare that reference access 40 percent situation with the same 40 percent scenario with open access on the left and limited access, you see kind of three interesting behaviors. The first is the impact that these siting restrictions can have on the amount of wind capacity needed to reach 40 percent.
In particular, look at the limited access scenario. There's 666 gigawatts of capacity to get to 40 percent, which is 100 gigawatts plus greater than under reference access. There's a lot of reasons for that. One is some of the high-capacity factor locations are no longer available.
So, you're relying on lower-quality wind sites. The second is on a curtailment of wind, and we'll get to that in the next slide. So, because we're applying this on a generation basis, the amount of capacity really varies depending on your level of siting restrictions.
Obvious from these maps is the plant locations vary significantly. Focusing on open access now, you can see that a lot of the wind development is still concentrated in the wind belt area, particularly with large power plants. The darker blue there compared to the yellow refers to more capacity per unit cell. So, you still see these large plants in the wind belt area. We do start to see some expansion particularly in the Appalachian regions, and you see smaller plants in the Great Lakes, Coastal, and Eastern Atlantic Coastal regions where those sites are available and are close to load centers.
So, become economically viable to hit 40 percent. Conversely, at the limited access, you see a very different picture for the future wind fleet that, again, hits the same 40 percent level nationally. First, all those blues are largely gone.
They're replaced with the yellows, meaning the plant sizes are much smaller. It's a little tricky here in our modeling where we had to break down the country in this gridded cell. Nonetheless, if you look at the median capacity that's developed each cell, it's only 46 megawatts. There's a lot of questions about economies of scale, balance of station cost that we're trying to tackle now.
We had it fully tackled in this analysis, nonetheless it indicates that the plant design may differ to get high wind sheers under the restricted siting regime. So, you see, obviously, the spread of wind across the country. You do see, however, that there's actually less deployment of wind into the high load pockets there. Those setback requirements are really preventing deployment even though they're very close to load centers and, therefore, could save you on saving transmission costs. Okay, so going back to that curtailment that I mentioned earlier.
Again, these are the same 3-40 percent wind scenarios for the three siting regimes. What obviously stands out here is the much higher level of curtailment under limited access. That's just simply because your outer resources in those zones where transmission is available or transmission is not too expensive and therefore, deploying in transmission congested regions where some of that wind power will need to be curtailed because of the transmission congestion. There's also a secondary effect on the generation profiles. You're not going to be able to tap into wind resources that have profiles that could be more easily integrated with the rest of the power system.
So, that's also a factor in our curtailment results. Of course, curtailment, the initial transmission expansion needs, and the exclusions of those high-capacity factor sites that I mentioned earlier can raise electricity costs and, in these cases, raise them pretty significantly by the 2050 time period. We're talking about $10.00, $15.00 per megawatt hour increase in these wholesale price metrics. So, because of the interactions with transmission here and some of the plant size aspects that we noted earlier, we did some sensitivities relative to this 40 percent wind limited access case.
One of them, in the dark red line with the highest level of curtailment on the left there, is one where we excluded new transmission, new interregional transmission as an option in the model altogether. As a result of that, you can see very high curtailment rates and very high electricity prices. The numbers are really kind of immaterial here, but really the most important message here is that when you have a regime that is difficult to site wind and difficult to build transmission, you're essentially going to be hard-pressed to extend wind generation beyond 30 percent or so. It's going to be really challenging.
Perhaps that the world we're in. Perhaps that's not going to be the world that we need to be concerned about. But that's certain what the modeling indicates that, if it is the world we're in, there's going to be some challenges there to further wind deployment. On the other side of the yellow lines there is one where we put in an HVDC macrogrid.
One big caveat with that is this macrogrid is not optimized for our specific scenarios. It's from other analyses from MISO and from the SEAM Study at NREL. Nonetheless, we see directionally that this macrogrid could reduce curtailment, could reduce prices. Some other follow-on work that we've done after this was completed indicates that the benefits can be even greater. So, that's something that we're pursuing is the ability to use transmission as a mitigation strategy and tap into wind sites that are available due to exclusions from other siting constraints.
The last sensitivity we ran is in the pink here, which we called a small site sensitivity. I didn't mention this earlier, but in terms of the data flow from reV to reEDS, we did exclude small sites with small available resource of less than 20 megawatts. Again, we haven't tackled economics of scale well in this analysis. So, we simply excluded them in most of the scenarios.
In this sensitivity, we put those back in. What we've noticed is, when you put those back in, a lot of those sites are prime sites in that they're close to load centers and therefore, they're accessed in this specific sensitivity. Again, the numbers maybe may not be too relevant, but what that tells us is that there could be some interest in designing wind plants, not just focusing on the large plants in Oklahoma, in Texas that are developed today. But instead, small scale wind power plants may be useful and valuable under this kind of constrained siting future. I'd like to summarize now with a few points.
The first point that we hope to have shown you is wind siting, especially for onshore wind can really impact the amount and location of wind deployment in the future, especially under scenarios with high demand for clean energy. We're talking impacts in the hundreds of gigawatts range and could become really meaningful. The second point here is we do notice that there is asymmetry between tightening and loosening the restrictions.
All this is to say is the marginal impact of restrictions depend on what regime you're living in. If you're already in an open access type of regime, slight changes to restrictions nationally at least – obviously, not locally – may not matter in the great scheme of things. However, if collectively we're in a constrained situation, an incremental restriction might be more problematic and could lead to higher prices and higher emissions from the power sector side.
Again, we need to consider the local aspect as well. The third point highlights some of those interactions between transmission and future wind plant design. We'll get to that in the future work slide next. But really, this is an area that is really important. I don't think our analytics and modeling have finally have caught up to the importance of these interactions. So, we're tackling some work in that space as well.
Overall, these findings really highlight the importance of considering some of these local land-use factors in regional energy planning or regional or national energy planning, and the very difficult challenge of weighing the local impacts to global impacts on from global climate change. One more slide here on the kind of future research. We recognize that or we think that what we've done has really advanced the prior work that we have done, particularly at NREL, but perhaps the broader research community here.
But of course, there's much more work to be done. So, we've grouped kind of the future work in three buckets. Some of this we're undergoing right now. I saw some questions about offshore earlier. That's one thing we're tackling. But others, we're going to need to continue to work on over the years.
So, these three buckets include considering additional siting factors. Other social impacts we haven't fully embedded; noise, flicker, visual impacts. Our setbacks are proxy to some of these but really haven't tackled them head on. Wildlife and ecological impacts are obviously very important and we're starting to get more of a handle on that.
Of course, other technologies. We focused on land-based wind for good reason. It has lower power density than other aspects. The siting challenges may be greater. However, there are nuances and aspects with solar, with transmission certainly, with nuclear power, et cetera that I think is worth looking into as well. Anthony talked about some of the impacts on turbine advancement.
You have bigger and bigger machines. Maybe there's more and more conflicts with those larger machines. But what are some of the other innovations that might overcome some of those obstacles or help alleviate some of these? That could include optimizing the plant design, optimizing the turbine layout, looking at hybrids, as well as looking at the operations of those plants. In terms of the wildlife space, one good example is curtailing the energy during times of bat or bird migrations. Lastly, on more of the energy systems side. We've looked at power system decarbonization.
It gets to the 95 percent level. That's certainly less than 100 percent emissions level, and it did not include increased demand for electrification, for hydrogen. A lot of work out there has showed a really high increase in overall electricity demand and, particularly, from clean energy resources like wind. So, taking those into account might even exacerbate some of the sensitivities to siting that we solve for. The last bullet I've already touched on which is the interaction between transmission expansion, the viability of that, the extent of that within the siting as well. We have several resources.
I think the top one we should put in the chat here. Eric, if you could do that. But we've put a lot of the data and resources and papers available, both at the supply curves that came out of Anthony's work that you can use for your own analysis as well as some of the underlying data that supported that supply curve work. For example, those 271 siting ordinances are available.
They were as of 2018. So, they will need some updating, but hopefully will be a good start. We do intend to update this website and build upon it when we conduct future research.
So, please, check in. If there's interest, please feel free to ping any of us. To conclude here, I do want to show a visualization that shows how the U.S. wind fleet has evolved over the past ten years, as well as might evolve under one of our scenarios through 2050.
It shows some of the themes that we've talked about before, which is an expansion geographically of wind development over time under this low carbon scenario. But it also shows here – well, it doesn't show. It's a single scenario.
But we hope our work shows that siting can be really important here in terms of what the future wind fleet might look like, what the future power system might look like, and the potential challenges with respect to siting and reaching kind of a low-carbon grid. So, with that, I think I will turn it back to you, Eric. Oh, actually hold on. Let me go one more slide here.
We forgot to thank some of our research collaborators in the very beginning but, certainly, we couldn't have done this work without their contributions. Travis Williams, Matt Mowers, Dylan Harrison Atlas, Billy Roberts, Galen Maclaurin, they all really contributed and helped to this body of work. They're coauthors in the two papers that were mentioned earlier. We'd also like to thank the Wind Office at the DOE for funding this support and particularly Patrick Gilman for his support and guidance and, of course, all of you for your interest here. But now, I think we have some time for Q&A.
Eric? Eric: Awesome. Thank you, Trieu. I think we've got a little over 15 minutes for Q&A. Our plan is to go to quarter past the hour here. There's been quite an active chat thread and there's actually been several questions answered in the chat thread.
Let's see here. I'm just trying to think of the most effective way. Maybe we can try and consolidate the conversation in the chat thread now to the active one here. So, I guess if you still have persistent questions, why don't we go ahead and either post them to the chat or open up your mic, unmute yourself, and go ahead and raise those with the raised hand function.
I think, Mary, you've got your hand up first here. So, why don't you go ahead. Audience: Hi. Thanks, both Anthony and Trieu. That was just so great.
I really, really enjoyed seeing how the information in your paper translated to the way that you had explained it. But just going to this energy justice theme, which is a huge theme for DOE right now and there have been a lot of questions back and forth in the text, I just want to elaborate – have you guys talk more about how this might possibly be incorporated into these types of siting studies at all phases because there is this social science side of the house and this technological side of the house, and we need to start looking at it from the very beginning, from the conceptual research throughout to the entire process. So, I'm just wondering if you could speak to that a little bit. Trieu: Yeah, thanks, Mary. I think that's one thing that I meant to mention at the very end there which is to extend this work really requires that collaboration that you highlighted there, not just from the wind technology experts, not just from the geospatial data scientists and the energy modelers, but to social scientists, to biologists, and to others. We started some of those collaborations and some of those folks are actually on this webinar.
So, that's great to see. In terms of the social energy justice type questions, certainly Anthony and I are not experts in this field. I think you can obviously layer on the geospatial aspects to some of these screening and dimensions and perhaps apply exclusions or costs or some considerations with relation to them. I think we are very interested in doing so, and we can imagine the framework that we've built for this can be extended to do so. But we'll need partners to do that. We don't have the capabilities ourselves.
Anthony, I don't know if there's more you want to add to that. Anthony: No, I think you said it perfectly. So, really just looking for collaborators on this kind of social dimension side of things.
But yeah, all the data is really there. We just need to make the linkages with other researchers and make it happen. M: Yeah, I think this could be a really great tool for informing a lot of future research in this space. So, hopefully the funders out there will hear that and see that. Trieu: Thanks, Mary. Eric: All right, we had one question that I don’t think got addressed in the chat, Trieu, and that is related to the curtailment numbers and the possibility or the applicability of that to potential storage solutions.
Maybe you could just say a little bit about how we treated storage and curtailment in the current work and how that might change going forward. Trieu: Yeah, good question there. So, the model does look at a variety of storage options.
We have battery energy storage from two to ten hours of energy duration. We have pump storage in there, and we have compressed air storage in there as well. All of those are in kind of the diurnal storage level. So, their ability to reduce curtailment, although modeled and captured in there, is going to be limited especially given the nature of wind production profiles, the kind of seasonal affects that you see.
What we're working on right now and maybe what the question is leading to is longer duration storage, hydrogen, other seasonal storage options. We are working on improving that right now and being able to capture the economics of that. I mean there's some cost considerations as well, but that's kind of ongoing work. Haven't fully built in. It's a valid point there. If you're constrained by transmission, constrained by siting, maybe these long duration storage options could save you in the end.
We don't know yet. Eric: Yeah, Trieu, just one other clarifying question and a comment, perhaps. When the model is building wind in these cases, you're looking at the economics, considering the marginal curtailment rates in a given area, but you're not providing the ability for the model to build the wind and utilize storage as part of the economic calculus at this stage.
Right? It's either you build storage, or you build wind? Trieu: The model does consider if you build wind and storage together you could reduce some of the curtailment on that. Again, the storage options are limited. The parameterizations may be somewhat limited as well. So, definitely that interaction between wind and storage is complicated. We've taken a first cut at it.
Needs room for improvement. Particularly, in other work that we've done, waiting till the longer duration storage for wind, in particular, which differs from how you might think about P.E. Eric: Sure. Then the comment is that to take this question one step further, it's actually pretty interesting to think about the relationship between siting and grid-disconnected or the potential for grid-disconnected wind deployments.
Obviously the resource is pretty far from the load in many cases, and our ability to build transmission or not is not yet clear. Obviously, the model takes a tact on that, but whether we can do that in the real world is certainly a significant uncertainty. However, if you could build a localized facility to generate electricity that's then converted into molecules or otherwise, it's a pretty interesting thought exercise, at least at this stage. A couple more questions. So, one of the first questions was about how the technical potential estimates that Anthony reported and then, I think, the scenarios, subsequently, that Trieu presented on, how that relates to some of the policy goals. There was some conversation in the chat about maybe we should be considering more aggressive scenarios given that there's a big emphasis on zero carbon in the electric sector by 2035.
Trieu or Anthony, do either of you want to talk any more about the relationship between this work and potential policy considerations? Trieu: Yeah, maybe I'll start there. We are conducting other work that looks at more aggressive carbon mitigation scenarios. So, that's underway; nothing that we can share right now.
The interaction with the supply parent work, I mean, I'll defer a little bit to Anthony here, but we are building on that. It's, as you can imagine, hard to get all these models to talk to each other or time intensive and whatnot. So, there are some constraints on our just time, real-world time to do these kinds of things. But certainly, they are interplayed there. I think the interplay with transmission policies is a really relevant one and we are looking into that at some extent now.
Anthony: Yeah, I also just mentioned on the chat there that we actually largely just wrapped up all of the modeling last fiscal year, kind of before kind of the current pushes for decarbonization, for rapid decarbonization. So, yeah, we are – as Trieu said, we are currently modeling these types of scenarios going forward. Just not in a position to share it just yet. Eric: The other thing that I would add to that perhaps is that, from a policy perspective, our philosophy generally is that we can do all of these things, but it matters kind of how things play out. So, how much wind do you need, where is that wind located, how much do you want to pay to accomplish these things? Those are sort of the relevant ways we look through it. But I think that there's still enough wind energy to meet very, very aggressive policy goals, but the precise pathways and the way we accomplish that or serve it is still really – provides a lot of really interesting research questions for us to dig into.
Gosh. We just got an additional question from the chat thread here about the transmission side of things. Trieu, do you want to elaborate any more on kind of how we're thinking about transmission going forward? Trieu: Sure. So, I see the chat here from David here on transmission policy. But overall, I think transmission modeling is an area that needed improvement here across the whole system. I think what we're doing – I can share what we're doing here is we're not fully capturing the capacity credit of transmission, if you will, and we're not adequately treating the ability for transmission to potentially use resources that are distant to provide capacity services to a region that needs that capacity.
Those capacity services can be very important under a low-carbon power system where you can't build local, clean, firm capacity but you may rely on transmission to do that. So, there's a lot of transmission modeling to do that. Maybe the one that gets more attention, and which is we're also working on, is looking at some of these macrogrid designs that are more optimized to our system. So, we are looking at that from a technical lens.
Before policy lens is looking at incentives to transmission or different cost of transmission might help in there. I realize that in the kind of economic folly world, it's somewhat limited and not as comprehensive as you might want to look at in terms of transmission interconnection policies across allocation questions and [inaudible]. In terms of broader scenarios to the question there, we're certainly looking into these 100 percent clean energy scenarios. We're looking into higher electrification scenarios.
We have already looked into higher electrification scenarios, but we need to extend that even further to look at kind of net zero energy system type of scenarios as well. Anthony: I'll just add, too, on the supply curve side. We don't apply the same kind of siting restrictiveness or rigor as we do for the wind plants for transmission lines, for the spur transmission lines.
So, we do have active work to start to integrate some of the social and environmental constraints associated with developing those spur transmission lines. That's ongoing work and we know that's something that we need to improve upon and capture better. Eric: We had an additional question here focusing on sort of the turbine technology and plant side of things. It's related to if you can get AEP gains from improved technology, how might that facilitate our ability to serve a particular policy, objective, or get a particular level of wind penetration. Maybe I'll just take a quick crack at this one.
I think that the short answer is, from a technocratic perspective, yeah, we can improve capacity factors for wind by one percent or whatever that equates to in terms of one percent change in AEP. We can run that through the model and see how that affects wind deployment. I don't think it would be a huge signal in the model in response to that. But I actually think the question is well-placed, because the relationship between plant sizes, plant costs, plant economics, and how that's used within the broad power sector from a continental perspective, that's actually a really interesting and rich question. There were some questions and comments in the chat about the plant sizes and, how you saw that in the more restricted scenarios, plant sizes were actually decreased.
So, what's the technology sort of maybe counter to that? How could the technology evolve