In this week’s episode, host Daniel Raimi talks with Dick Schmalensee, a professor emeritus of the MIT Sloan School of Management and a former chair of the board of directors at Resources for the Future. Schmalensee and Raimi cover the takeaways from a recent study on the future of energy storage that Schmalensee coauthored, along with related insights and implications for current and future policy. They discuss the role of energy storage in a net-zero-emissions electricity system, the strengths and weaknesses of key energy storage technologies, and what these technologies might cost.
Listen to the Podcast
- Efficient energy storage complements the transition to renewables: “As we decarbonize the electric power sector and hopefully the rest of the economy, most plans call for very heavy increases in the use of wind and solar generation. Wind and solar generation are lovely, but they’re intermittent—that is to say, their output varies over time. And it’s imperfectly predictable. Storage plays a potentially huge role in systems that are dominated by wind and solar generation because it in effect moves generation from one time to another.” (2:45)
- Grid flexibility through expanded storage and transmission capacity: “If you have long-duration storage available at the kind of costs that people forecast, you use it. It substitutes for lithium-ion, it substitutes for generation—basically, everything substitutes for everything else. Transmission substitutes because good wind and solar sites tend to be remote. So, if you can build transmission, you lower costs. Storage substitutes for generation, and long-duration substitutes for short-duration.” (17:43)
- Option of turning old coal plants into experimental storage facilities: “It’s sort of important now, before we tear all these old plants down, to use one or two of them as a pilot project to show the potential, if it’s there, of this technology—of reuse of coal-fired plants as storage centers, or storage installations, for what would be long-term storage.” (28:09)
Top of the Stack
- “The Future of Energy Storage” by Robert Armstrong, Yet-Ming Chiang, Howard Gruenspecht, Fikile Brushett, John Deutch, Seiji Engelkemier, Emre Gençer, Robert Jaffe, Paul Joskow, Dharik Mallapragada, Elsa Olivetti, Richard Schmalensee, Robert Stoner, Chi-Jen Yang, Bjorn Brandtzaeg, Patrick Brown, Kevin Huang, Johannes Pfeifenberger, Francis O’Sullivan, Yang Shao-Horn, Meia Alsup, Andres Badel, Marc Barbar, Weiran Gao, Drake Hernandez, Cristian Junge, Thaneer Malai Narayanan, Kara Rodby, and Cathy Wang
- “As the Great Salt Lake Dries Up, Utah Faces an ‘Environmental Nuclear Bomb’” by Christopher Flavelle
- “A durable US climate strategy … or a house of cards?” by Richard Richels, Benjamin Santer, Henry Jacoby, and Gary Yohe
The Full Transcript
Daniel Raimi: Hello, and welcome to Resources Radio, a weekly podcast from Resources for the Future (RFF). I’m your host, Daniel Raimi.
Today we talk with Dick Schmalensee, Professor Emeritus of Economics and Management at the MIT Sloan School of Management and a former board chair here at RFF. Dick was a coauthor on MIT’s recent Future of Energy Storage study, which assesses the role that energy storage might play in a net-zero-emissions electricity system. In today’s conversation, he’ll help us understand the key technologies, their relative strengths and weaknesses, and how they vary in terms of costs. We’ll discuss key results from the study, some policy implications, and how today’s policies can best support the technologies we need in the decades to come. Stay with us.
All right, Dick Schmalensee from MIT, welcome to Resources Radio.
Dick Schmalensee: It’s a great pleasure to be here.
Daniel Raimi: So, Dick, we’re going to talk today about a report that you contributed to on the future of energy storage. But before we do that, we always ask our guests how they got interested in working on energy or environmental issues. So, how did you come to this field?
Dick Schmalensee: It’s funny. Among my influences in undergraduate and graduate school were Mo Adelman, famous for work on oil, and Paul McEvoy, less famous, but deservedly famous for work on natural gas. But they didn’t capture my interest. It really happened much later when I came back to MIT in 1977. And there were just a lot of people doing interesting work on energy-related issues. And I kind of got caught up in all that. And my colleague, Paul Jaskow, kind of pulled me into work on electricity. And then later on, really, after some time in Washington, I became fundamentally concerned about climate change, which of course gets you immediately into energy/environmental issues. And that’s been a driver of my work for some years.
Daniel Raimi: Yeah, that’s great. I mean, so many people who we have on the show come to energy and environment in all sorts of different ways, sometimes early in life, and sometimes later.
As I mentioned, we’re going to talk about this Future of Energy Storage report from MIT. And it’s one in a series of reports that MIT has issued over the years about the futures of different energy technologies. To get us started, can you help us understand why energy storage is likely to play a much larger role in our energy system in the future—especially as we think about decarbonization?
Dick Schmalensee: Happy to do it. And this was a real driver of the study. As we decarbonize the electric power sector and hopefully the rest of the economy, most plans call for very heavy increases in the use of wind and solar generation. Wind and solar generation are lovely, but they’re intermittent—that is to say, their output varies over time. And it’s imperfectly predictable. Storage plays a potentially huge role in systems that are dominated by wind and solar generation because it in effect moves generation from one time to another. Buy low, sell high means you charge storage when there’s a lot of energy and a lot of generation output, and you bring it to the system when there isn’t enough. So, storage now doesn’t have much to do, but in the future it will have a lot to do.
Daniel Raimi: That’s great. And you and your coauthors in the report discuss a pretty wide range of technologies, and we’re not going to have time to get into all of them. And we would certainly encourage people to read the report, to get all of the detail that’s in there. But can you give us a thumbnail sketch of some of the types of technologies that you consider in the report and how they stack up in terms of their ability to do things like store power and release power, as well as how they stack up in terms of their costs?
Dick Schmalensee: Well, the cost is the key. And we (by “we,” I mean, of course, the technical people who worked on the report) consider a wide range of technologies and proposed technologies, and people with an interest in technology should look at those chapters. The technologies that seem plausible to play a role in 2050, which was our focus, come in three groups, basically.
The first group is lithium-ion batteries, and lithium-ion batteries are very good in terms of cost at expanding how much power they can deliver at any instant. You need that in electric vehicles, of course. But expanding the length of time over which they can deliver that power is expensive relative to alternatives.
Now, the other group I'd say is typified by pumped-hydro storage, or pumped-storage hydro as it’s termed, oddly enough. It’s relatively easy to expand the duration, the length of time, for which the system can deliver power, because you’re basically expanding the size of the uphill reservoir. That’s comparatively cheap. What’s expensive is delivering power. You’ve got to build the turbines and the tunnels and all of that stuff. So, something like pumped-storage hydro and the use of hydrogen comes in there. And some thermal-storage technologies that come in that category are relatively cheap at being able to provide duration—being able to provide energy for long periods. But it is expensive in terms of providing power at any instant.
Then there’s an intermediate group, typified by so-called flow batteries. That’s kind of intermediate on both dimensions of cost in terms of the cost of power and the cost of storage.
I should say that storage is much more complicated than generation. Even a simple representation of storage takes something like seven parameters. I describe two of them, two cost parameters, but there are, in principle, seven of them, and it’s much more complicated than describing generation technologies. And there are a wide range of things that people talk about, like compressed air, that our technical people say, “No, that is just not going anywhere.” It occurs a lot in casual discussions because they built two plants, but it doesn’t seem likely to go anywhere. So, it’s discussed in the technical chapters, but not in the modeling chapters with which I was mainly involved.
Daniel Raimi: I know you’re not focused on the technologies, but I think it might be helpful for our listeners to get just one level deeper on a couple of these technologies. You mentioned pumped hydro a couple times, as well as hydrogen. For those folks who don’t know how pumped hydro works, can you briefly describe just what’s involved in using it? And also for hydrogen: what are the different ways that you considered to produce hydrogen in this study?
Dick Schmalensee: Sure. Pumped hydro is an old technology. It may date back to the ’20s, or maybe even earlier. It was deployed here and in Europe, when it was decided in the early ’50s that nuclear power was going to dominate. And the good thing about nuclear is the running cost is cheap. A bad thing about nuclear is it doesn’t easily vary output to respond to changes in demand. So the idea was that we’d use pumped hydro when generation would exceed demand—say, at night. We’d use that excess power to pump water uphill from some low reservoir to some high reservoir. And then, during the day, or at other times when demand exceeded nuclear output, we would use that water just in an ordinary turbine to generate electricity to supply to the grid.
There are facilities operating in the United States. There are facilities operating in Europe. China is building a lot of them. Nobody else is. China is building a lot of them. They don’t make a lot of sense now in the United States because we have natural gas plants that are very good at following supply-demand imbalances. So we don’t need much pumped hydro in the United States now, although there are proposals to build some.
Hydrogen is interesting. Hydrogen comes in various colors that people talk about. It’s mostly produced now from natural gas. What we focused on, which we think is relevant down the road in mid-century, is hydrogen produced by splitting water by electrolysis, because you can do that with solar power or wind power. So, you can produce hydrogen with no carbon emissions. And it’s cheap to store. It’s cheap to store in tanks. It’s even cheaper to store underground, so you can produce a lot of it and then use it over time.
So, you can produce weeks’ worth of hydrogen in storage, and expanding the storage is relatively cheap. The problem is with foreseeable changes in technology. That’s going to be expensive. So, it turns out not to be optimal to use hydrogen for long-duration storage, which is what it’s good for. Relatively inefficient—that is to say, you lose a lot of energy in going from electricity to hydrogen, and then back to electricity. That’s so-called round-trip efficiency.
You lose a lot, but here’s the kicker: We explored a couple of studies reported in the future of storage study and in other publications. Colleagues reported experimenting with, well–suppose you used hydrogen to replace natural gas in industrial uses. And the modeled area is Texas, where there is a lot of natural gas used for heat. “Suppose,” they said (I wasn’t involved in the nuts and bolts of these studies), suppose you could use hydrogen instead. Well, you could store the hydrogen underground, or you could store it in tanks. Either way, it’s pretty cheap to store. And then you can generate hydrogen by electrolysis. So you’re mostly using it to replace natural gas. But what you’re doing is making the production of hydrogen a flexible demand. So, when you are short of generation relative to demand, you produce less hydrogen. When you are long on generation relative to demand, you produce a lot of hydrogen. Well, it’s the flexibility in demand that sort of replaces storage. You’re not using the hydrogen cycle as just a storage medium. You’re using it as a flexible load to produce hydrogen for industrial use.
That turns out, in at least some regions, to have a lot of promise. And again, there are underground caverns. You can store enough hydrogen to generate electricity for long periods of time, which is pretty useful, because there are times when the wind just doesn’t blow or it’s cloudy for a long time. Lithium-ion batteries don’t make a lot of sense beyond a few hours of duration. So, we find that very interesting—the use of hydrogen beyond electric power as a flexible load, as they describe it, to help deal with intermittency of generation.
Daniel Raimi: That’s really interesting. Thank you for all those descriptions. That’s all really, really helpful.
So, again, recognizing that we’re glossing over tons of detail here, I’m hoping we can jump to some of the key results and the implications of those results that you and your coauthors find from some of the scenarios that you model out into the future.
Dick Schmalensee: Sure. Let me describe roughly what we did and then come to what we think it implies and some of the puzzles it poses. So, the technical people gave us high, medium, and low forecasts for the costs on seven dimensions—seven parameters for all these technologies in 2050. Now, the National Renewable Energy Lab has some cost forecasts, and our people produce some independently. So, here we have forecasts of storage technologies around mid-century. And we say, “Well, what role will they play in the power systems at that point?” And to do that, you sort of have to model power systems at mid-century. So we have projections for demand, assuming there’s a lot of electrification, and we have cost projections for wind and solar generation and for natural gas, with carbon capture and sequestration, where you put the CO2 underground.
So, we looked at three US regions. We looked at Texas, we looked at the Northeast (New York and New England), and we looked at the Southeast (basically Georgia, Florida, Alabama, Mississippi—most of that). And we said, first question: If those systems are efficient in 2050, what do they look like without constraints on carbon emissions? What do they look like? Well, the answer is they’re very different. Texas, even with no pressure to reduce CO2 emissions, would use a lot of wind and solar generation, because Texas has good wind resources, as we know, and it has sunshine. So, even without any constraints on carbon emissions, Texas would use a lot of wind and solar. And as a consequence have a lot of storage.
In the Northeast, where I live, we do have wind, but it’s very hard to put wind onshore because of the density of population. Offshore wind is not cheap. And there’s not a whole lot of sunshine relative to other places. So, if you just let the system alone, we model that in 2050 it would in fact become more carbon intensive because we think the nuclear capacity operating in the Northeast would shut down and be replaced by natural gas. In the Southeast, we think that some nuclear reactors would stay in place and, though it doesn’t have great wind, it has good sunshine and would, in 2050, all else equal, without policy changes, become less carbon intensive.
Then we imposed carbon-emission constraints. We said, okay, what would these systems look like, and what would storage do in these three regions if, let’s say, we start imposing carbon-emission constraints? And the way you do that in modeling is basically a carbon tax, or a carbon charge. That’s the easiest way and, second of all, the cheapest way, as we all know, to reduce carbon emissions.
So, we do that in each of the three regions, and we use different assumptions about costs and technologies. A first result is that you can take almost all the carbon out of these three systems without huge increases in costs and without really compromising reliability, even if you have only lithium-ion batteries at your disposal. And the reason that works (and I say “almost all”) is that it turns out to be economical, on our assumptions, to retain natural gas capacity for times when the sun isn’t shining and the wind isn’t blowing for long periods of time. So, you have natural gas capacity, and you don’t use it much. To go all the way to zero emissions means you have to replace that natural gas capacity. And it’s very expensive to go that last few feet to zero emissions. So, going to near zero turns out to be relatively inexpensive.
I want to say a little bit in a second about how you do it, which is interesting in its own right. But another result is if you have long-duration storage available at the kind of costs that people forecast, you use it. It substitutes for lithium-ion, it substitutes for generation—basically, everything substitutes for everything else. Transmission substitutes because good wind and solar sites tend to be remote. So, if you can build transmission, you lower costs. Storage substitutes for generation, and long duration substitutes for short duration. So, I hope that makes sense as a rough overview.
Daniel Raimi: Yeah, that’s great. Can I actually just dig into one of those topics real quick?
Dick Schmalensee: Yeah.
Daniel Raimi: So, assumptions about transmission: All of the studies that I’ve seen in recent years come to very similar conclusions to what you described–that we can get to near zero at relatively low cost in the electricity system. But my sense is that most of these studies assume that we are able to sort of build transmission wherever and whenever we want. So, I’m wondering how—if that plays an important role in the modeling that you carried out—how well you think that might actually reflect reality.
Dick Schmalensee: Well, it did play an important role, and it varies by region. Texas has radically expanded its transmission capacity internally to bring wind to load centers. And it seemed a reasonable approximation for 2050 to assume that Texas had fully built out its transmission system and could be treated as sort of one zone. In New England and New York, not so much, because it is very hard to build transmission into New York City. If you don’t take into account the difficulties of building transmission and the difficulty of siting wind resources, the system, to reduce cost, tends to want to build massive amounts of transmission into New York City and to put a lot of wind onshore. We basically constrained the system not to put wind onshore, because you just can’t do it in this region. And we limited the amount of transmission that can be built within the New York/New England region.
We assumed, by the way, that there would be hydropower available from Quebec, despite the recent enormous difficulty of building a line, but we didn’t assume much expansion. We already get hydro from Quebec. We didn’t assume expansion. In the Northeast, when you really clamp down on the carbon constraints, the system wants to build transmission to link zones with a lot of wind and solar resources to zones that have less and have a lot of load.
We think we modeled it reasonably. The Southeast is intermediate between these two cases. Related studies that we cite that were done as part of our report and published before it look at more ambitious transmission expansion—a national grid, so to speak—that would link Texas and the West and everybody else. And again, that study and other studies show great cost-reduction potential from that transmission. Now, whether you can build it or not, is, at the end of the day, a political question. I mean, it’s an intensely political question in the Northeast to bring hydropower from Quebec into, well, particularly into Massachusetts, because we have aggressive decarbonization goals here. And the cost of meeting those goals without hydropower from Canada is going to be very high. But voters in Maine have blocked the essential line. And we may do exotic things, like undersea cabling. It’s not entirely clear how it can be done.
Daniel Raimi: Great. I sort of got you off track there with that transmission discussion. So please keep going on some of the key results and implications on storage.
Dick Schmalensee: Well, one of the findings that is very robust and was at first surprising, and then not so much once we thought about it, is that as you tighten the carbon constraint, the wholesale price of electricity is much more often at or near zero. Now, that’s a puzzle, right? Because the average price is going up as you impose tighter constraints on carbon emissions. But one of the responses to a tighter constraint on carbon emissions is that you build more storage, of course, so that you can take generation into time periods when there is less wind and less sunshine. But another response, which you compare on the margin, is to just build more generation, right? You can deal with cloudy days either by storing energy on sunny days and using it on cloudy days, or by building enough solar capacity that you’re okay on cloudy days.
And it turns out it’s optimal to do both. When you build more wind and solar generation capacity, the fraction of hours at which you’ve got excess generation goes up. And what that means is that you have a lot of zero-price days that we don’t see now. We don’t see that, and that’s a big deal. And it’s a big deal, potentially, at the retail level. If you want to electrify the economy and decarbonize other sectors by electrifying them, you want users to use electricity when it’s free.
The story I tell all the time is about my son in Hawaii who has an electric vehicle, and he pays something like 30 or 35 cents a kilowatt-hour to charge that vehicle—even when the Hawaiian system is desperately trying to get rid of excess solar generation. Now, that’s crazy, to use a technical term. And it’s going to become a more widespread problem.
So, what we uncovered, and what a number of us are scratching our heads about, is: How do you do retail rates? Because along with all those hours of near-zero wholesale prices come more hours of very high prices, right? The average has to go up. So, if you have a lot of zeros, you’re going to have a lot more high prices. Consumers should see the low prices; that’s true. But as we saw in Texas, telling consumers they should also see the high prices is just not on.
So, how do you provide insurance? How do you provide hedges? How do you come up with rates that encourage electrification and economy-wide decarbonization, without imposing unbearable risks on ultimate consumers? I think that’s a puzzle. I think it’s soluble, but it’s not a simple puzzle.
Daniel Raimi: That’s so interesting. I would love to actually do a whole podcast just on that question alone. It seems like there’s a million interesting things to dive into there.
Dick Schmalensee: I wish we had dived a little more deeply, because I don’t have a half an hour’s worth of answers on that question—but give us time.
Daniel Raimi: Okay, great. We’ll have you back some other time to talk about it.
So, we are nearing the end of our discussion, and I want to make sure to ask you a policy question. We’ve had a lot of policy implications, but a direct policy question would be: When you think about policies that government actors could take to incentivize a deployment of storage, we have different types of storage, right? We have some technologies that are quite mature, like lithium-ion, and we have others that are much less mature. Like, let’s say, metal-air batteries, or some of these green hydrogen technologies. In your view, what’s the right mix of policies that we should be thinking about when it comes to incentivizing mature technologies versus incentivizing less mature technologies?
Dick Schmalensee: Well, lithium-ion development is being driven by the electric vehicle business. Any reasonable projection of the use of lithium-ion in electric power systems is going to be dwarfed by what happens in electric vehicles. Manufacturers of vehicles and manufacturers of batteries have enormous incentives now to make lithium-ion better. I don’t see a federal role there at this point, other than to encourage the continued deployment of electric vehicles. Other technologies are where we think research and development efforts—and demonstrations, for that matter—should focus. Particularly long-duration technologies, like those involving hydrogen or some thermal technologies that are fun to talk about but really haven’t been done at scale.
The one exception to just R&D on technology development is using old power plant sites. Take a coal plant, and we’re not going to use it to burn coal anymore, but once you decide not to do that, you’ve still got the connection to the grid. You’ve still got the turbines; you’ve still got the boilers. And suppose you take generation off the grid and use it to heat something—rocks. There are a lot of rocks in this world. Use it to heat the rocks. They stay hot for a while, if you do it appropriately. And then you use that heat to generate steam, to produce power in what used to be a coal-fired plant. A number of folks in our study explored this possibility and think it has promise. If that’s right, then it’s sort of important now, before we tear all these old plants down, to use one or two of them as a pilot project to show the potential, if it’s there, of this technology—of reuse of coal-fired plants as storage centers, or storage installations, for what would be long-term storage. So, deploy that now as a pilot plant.
For some of this more exotic stuff, we’re still in the lab-plus-demonstration area. And we should be. There’s no point in rushing development. Something like pumped hydro, it’s a pretty well-understood technology. Deploying it now makes no sense. Power prices are relatively stable. And until we have a lot more wind and solar, there isn’t going to be the buy-low, sell-high opportunity that we see in our modeling for 2050. So, there’s no point in encouraging the deployment of pumped hydro and the other technologies that really aren’t ready. We need pre-commercial research development, and in some cases, demonstration. Notably, this old coal plant conversion opportunity needs to be demonstrated.
Daniel Raimi: It’s really interesting that, for that technology in particular, you could imagine it being appealing to communities that are struggling from coal plant closures, with jobs and tax revenue and stuff like that.
Dick Schmalensee: It would have all of that. It would have fewer jobs, I think, than attending to a coal plant, but it would have jobs, and it would certainly have property tax implications.
Daniel Raimi: Yeah, absolutely. Well, Dick Schmalensee from MIT, this has been a fascinating conversation. And, as we’ve said a couple times, we’re really just scratching the surface here. So I’d encourage people to check out the full report, which we’ll link in the show notes of this conversation.
And we want to close the conversation with the same question we ask all of our guests—to recommend something that you’ve read or watched or heard that you think is great and that you think our listeners might enjoy. So, Dick, what’s at the top of your literal or metaphorical reading stack?
Dick Schmalensee: Well, I’ll just mention two things I saw this morning. The first was a piece in the New York Times about the impending environmental catastrophe associated with the shrinkage of the Great Salt Lake in Utah. Really quite, quite astonishing and disturbing as part of the drought or as one of the consequences of the drought that has engulfed the West for some time.
And the second was an opinion piece in a publication put out by Yale. I don’t remember the name, but a number of my MIT colleagues make the point that we’re all consumed right now with worrying about the war in Ukraine and its implications for economic growth and inflation and everything else, which kind of takes our eyes off the climate problem. The climate problem persists—the climate problem really is an existential problem for the species, more or less. And yeah, the war is important and yeah, we need to pay attention to it. But if we take our eye off the climate problem now, the consequences will be potentially disastrous.
So, those two things caught my eye this morning. I think both are important.
Daniel Raimi: Yeah, absolutely. And just so folks know, we’re recording on June 8, so the recommendations might be a week or two out of date by the time you hear them, but we will have links to them. And I’m sure the content will still be relevant, whether you’re listening on June 15 or later in the month or anytime during the summer.
So, once again, Dick Schmalensee from MIT, thank you so much for coming onto Resources Radio, helping us understand the technologies and the key results that you and your colleagues have come up with, with the future of energy storage.
Dick Schmalensee: It’s been a great pleasure. Thank you for having me.
Daniel Raimi: You’ve been listening to Resources Radio, a podcast from Resources for the Future, or RFF. If you have a minute, we’d really appreciate you leaving us a rating or a comment on your podcast platform of choice. Also, feel free to send us your suggestions for future episodes.
This podcast is made possible with the generous financial support of our listeners. You can help us continue producing these kinds of discussions on the topics that you care about by making a donation to Resources for the Future online at rff.org/donate.
RFF is an independent, nonprofit research institution in Washington, DC. Our mission is to improve environmental, energy, and natural resource decisions through impartial economic research and policy engagement. The views expressed on this podcast are solely those of the podcast guests and may differ from those of RFF experts, its officers, or its directors. RFF does not take positions on specific legislative proposals.
Resources Radio is produced by Elizabeth Wason, with music by me, Daniel Raimi. Join us next week for another episode.