There is not much argument now about whether the United States can get to a 100% carbon-free electricity sector in the next 15 years or so. But many still believe that nuclear energy will be needed for the job as a complement to wind and solar. Their arguments center on the following:
1. It will take too much storage to make up for the variability of solar and wind. Therefore, dispatchable sources are necessary; in particular, nuclear is necessary.
2. Solar and wind take up large areas of land.
It doesn’t take a very elaborate technical analysis to conclude that solar + wind + battery storage is a critical but partial answer to the variability of wind and solar. But the claim that solar and wind therefore need nuclear as a complement derives from a failure to examine the full array of technologies already at our disposal. I’ve spent a few years looking at this issue in depth. The analysis shows that a 100% renewable electricity system based on solar and wind would be economical and also be more reliable and resilient in the face of climate extremes than the electricity system we have today, where centralized nuclear, coal, and fossil gas (aka “natural gas”) plants are the mainstays of supply. It would also be a lot cleaner.
You can find most details in a 2016 300+ page IEER report (Prosperous, Renewable Maryland). I’ll give a sketch in this blog and add some new details. We (IEER staff) did hour-by-hour modeling that included the supply needed to electrify transportation and space and water heating, so that CO2 emissions could be reduced from those sectors as well. Efficiency is an excellent starting point, making the system more economical; we factored that in first. We downloaded hourly solar and wind data (onshore and offshore) for various locations and combined them to get an hourly primary supply picture, along with the small amount of existing hydroelectric power (about 2%) in the state. We obtained utility demand data for Maryland and made hourly estimates of individual major components, including space heating, air-conditioning, water heating, and clothes washing. Our modeling of lighting took seasonal variation into account. Then we added battery storage. Here is what we found:
1. The system worked best when solar and wind generation were roughly balanced on an annual basis – wind supplies some electricity at night, when, by definition, there is no solar. Wind is also more plentiful in the winter, making it a good seasonal complement to solar.
2. Solar, wind and hydro and a small amount of industrial combined heat and power (using renewable hydrogen) met all of the load for 68% of the hours of the year, and varying amounts of load for the rest of the hours. In the lowest supply hours, the fraction of the load that was not met considerably exceeded 50%.
3. Adding just 5.5 hours of battery storage (average hourly load) increased the hours of meeting all the demand from 68% to 96%. Much, much better but still not good enough. Trying to increase battery capacity to meet the rest of the load makes it quickly apparent that the needed capacity spirals to huge, impractical numbers.
That is where the “smart grid” comes in. In a smart grid, communications go from one way — from the consumer, who flips a switch “on” or “off,” to the utility, which supplies the electrons instantly — to two ways: supply and demand can “talk” to each other, enabling what is known as “demand response” – the ability to adjust load to supply, if needed. The Federal Energy Regulatory Commission (FERC) has recognized demand response as a resource for the electricity grid (FERC Order 2222).
During hours when the solar-wind-battery combination does not meet the load fully, the smart grid can work with smart appliances (already on the market) to defer the unmet load to some other hour of the day for consumers who have chosen that option; they would get paid for signing up. FERC Order 2222 recognizes the technical equivalence of demand response, from the grid operation standpoint, to increasing the output of an electric power plant.
How would it work in everyday life? The hours when the solar-wind-battery combination does not meet load generally occur on days when there are other hours with surplus supply; this means demand can be deferred to some other time within a day when surplus solar and wind are available. Here are two examples.
Washing clothes: You load the machine and press “Start.” It happens to be a time when the grid has no available solar-wind-battery supply. For a rebate on your electricity bill, you have pre-selected the option of deferring your clothes washing to some other time of that day, should the need arise. Most of the time, the washing machine will start right away; but a small number of times, when supply is short, your clothes washing will be deferred automatically to a time within a 24-hour window when there is excess supply. You get an additional rebate on your electricity bill when the grid actually defers your demand. But suppose you are in a real hurry this time. You’ll have an override button, and your clothes washer will come on right way; you’ll pay more for that wash. (I’ll soon tell you where that electricity will come from.) If you don’t sign up at all for demand response, your clothes will always be washed when you press “Start” and you’ll always pay more when the grid has the most demand relative to supply. That is how peak electricity pricing works even today; it will be the same thing but in a new context, more nuanced, and more widespread. (That is also why renters, especially low-income renters, need to be protected during the transition to a smart grid: they don’t control what appliances they have. More on that in a future blog.)
Charging an electric car (by the time a grid is fully renewable, most cars will be electric): You plug it in to your home charger; you need 50 miles more of charge as soon as possible. The car starts charging and will be done in, say, two hours, independent of the state of the grid. You would pay more for that kind of on-demand charging, though, in a well-designed system, it will still be cheaper per mile than gasoline. But say you plug in at 10 p.m. and don’t need the 50 miles of added charge until commuting time at 7 a.m. the next morning. Your car will charge automatically for two hours but over the nine-hour window at times when there is surplus supply; as a result, you’ll have a cheaper commute. The potential for demand response is even greater when it is aggregated across consumers — something a smart grid would also enable. Allowing aggregation of demand response and other distributed resources is a principal feature of FERC Order 2222.
Given the attractiveness and flexibility of demand response, most people will sign up to save money. But what happens when people override their choices? That will be figured into the grid reliability calculation. The grid depends even now on what is called a “diversity factor” — not everyone cooks or bathes at the same time. Not all refrigerator compressors come on at the same time; neither do the electrical heating elements that keep freezers defrosted. Similarly, only some people will want to override a preset choice at any particular time. The grid won’t care which ones they are; a diversity factor, with a safety margin, will be built in.
Because some surplus electricity is available at some hour of essentially every day, a good bit of the demand that batteries cannot meet is now met by renewable generation but not at the exact time when the “Start” button is pressed. (Of course some things are not suitable for demand response, notably lighting; that is built into the design of demand response.)
These examples demonstrate a crucial fact: demand response reduces the cost of electricity to consumers because they get paid for contributing to grid resources. If the investment is only on the generation side, as for instance in peaking generation, then consumers pay all that added cost for resources that are used only a tiny fraction of the time.
Even after demand response there may be a tiny balance of the annual load for small fraction of the year’s hours. It was between one and two percent in the Maryland example we analyzed. The technology do meet that demand is also available.
Probably the best way would be to use “vehicle-to-grid” (V2G) technology. With EVs the same plug that charges the vehicle can also, with the appropriate technology and software, feed electricity to the grid — that is the “to-grid” part of V2G. It’s been developed and tested; it works. But can it meet the load? Is it comparable to nuclear plants?
Take long-term parking lots at airports. With suitable equipment, each could function as a sizable power plant. For example, the Thurgood Marshall Baltimore-Washington International Airport has more than 10,000 long-term parking spots. If three-quarters full, the cars parked and plugged in could supply the power equivalent of a large nuclear reactor for a short time (which is the nature of the residual peak load need). The daily lots could supply more. And that is just one airport! You park, you set how much minimum charge you must have when you arrive back. And you get a discount on your parking. If the grid actually uses your car battery, you get paid to park your car. An overview number: the combined horsepower of today’s vehicles is more than 50 times the combined capacity of all electric generation stations. Do we really need new nuclear, which makes plutonium just to boil water and consumes an enormous amount of water, in a world with security problems and climate extremes? Here is a recent article on that topic by M.V. Ramana and me, done for the Environmental Working Group. Rather, shouldn’t we we be thinking of saying bye-bye, as mindfully as possible, to the era of “Atoms for Peace”?
There are many other advantages of V2G. For example, when most or all heating is electrified efficiently and solar and wind are the main energy supply, peak demand will tend to occur on cold, windless winter nights. That is when school buses are parked. School districts could make money by signing up their electric buses up for V2G. Ditto for many other vehicles. Lawn care companies could make money by lending their battery powered mowers machines to the grid at night.
Another way to meet residual peak demand would be to make hydrogen when there is surplus supply of solar and wind. Hydrogen production using electricity (by electrolysis of water) is a form of energy storage; it can be done at power stations. (Large power stations today don’t produce hydrogen but they do use it to cool the electric generators, which enables operation at the maximum possible efficiency. It is stored on site; it is a familiar material in the electric power business.) When needed the hydrogen provides the fuel for light duty fuel cells of the type that are now used in fuel cell vehicles. This is the approach we modeled because it was simpler to do so.
Every single technology needed is available now to enable a transition to a resilient, reliable, democratized, economical, and 100% renewable, clean grid in 15 years. I would go so far as to say that demand response (including V2G) is the present-day equivalent of the post-1973 energy crisis understanding that energy efficiency could do the same things as energy supply, only more cleanly and cheaply. (For a personal account of that, see my tribute to the pioneer of energy efficiency policy, Dave Freeman, who passed away last year; he understood the role of efficiency well before 1973.) Unfortunately, much modeling today does not include demand response integrally.
Now, a note on land area: Fossil and nuclear power plants need fuel, which means more land every year for mining that fuel. Nuclear fuel is compact by the time it reaches the reactor; but every ton of reactor fuel requires on the order of a thousand tons of uranium ore (give or take, depending on the quality of the ore). There are hundreds of millions of tons of uranium mining and milling wastes in the United States, mainly on the Colorado Plateau, from nuclear weapons and nuclear power, despite the fact that the United States has been importing most of its nuclear power uranium requirements for decades — creating wastes in other countries like Canada, Australia, and Kazakhstan. Coal, petroleum and fossil gas also use land for fuel production and pipelines, not to speak of the area of flattened mountaintops, contaminated streams, and coal ash ponds.
Consider also this statistic on land use: today, about 30 million acres of agricultural land are devoted to the production of ethanol from corn, mostly for 10% of automotive fuel. This is far more than wind and solar land requirements. Here is a thought experiment. Suppose all the electricity the United Stats uses were supplied by ground-mounted solar. It would take roughly 10 million acres, or one-third of the land now used for corn ethanol. Further, the construction footprint of that solar – the steel to hold up the panels and the concrete footings would be at most 200,000 acres and probably much less. (The construction footprint of wind is also small.) Almost all of the the rest of the area could be used to grow food or graze sheep or return it to native grasses to enrich the soil and put back carbon in the soil.
Of course, we would not build a renewable energy system with only ground-mounted solar; far from it. A balanced renewable energy system will have onshore and offshore wind; it will have rooftop solar (for solar on new homes, see my 2020 report Gold on the Roof), urban ground-mounted solar, solar in parking lots, and solar on brownfields. Whenever solar is constructed on farmland, the area could be used to help join the food and energy systems so as to make them both healthier, more resilient, and more sustainable, while economically strengthening family farms and ranches. (See my 2021 report Exploring Farming and Solar Synergies)
Does renewable energy have environmental impacts? Yes. The scales of mining and construction impacts of a large power systems are roughly comparable — they all require a large amount of construction intensive investment, though specific materials and impacts depend on the technology. Where solar and wind shine (so to speak) is that they need no fuel. Rather, Mother Nature provides free fuel; we invest in the technology to harness it. To minimize the impact, whatever the energy system, it’s best to conserve energy and use it efficiently. We’ll be even better off and reduce mining impacts if we put in place facilities to recycle solar panels and batteries at the start of our renewable energy journey. That too, I’m saving for another blog. This one is already long enough.
A note of thanks to the Town Creek Foundation, which funded IEER’s Maryland work in its entirety for several years; the foundation made its last grants and closed its doors at the end of 2019.