August 21, 2020
Why Nuclear Is Not the Answer
Nuclear power, developed in the 1960’s, promised a world in which electricity would be “too cheap to meter”. This glorious future appeared tantalizingly within reach in the mid-1970’s, when the world was in a nuclear building boom. 50 large reactors would start construction every year, each capable of producing 1 gigawatt (a billion watts) of electricity.
Output of power is the size of the plant multiplied by hours of production, so a one gigawatt nuclear plant running for 24 hours would generated 24 gigawatt hours, and for 365 days would do 8,760 gigawatt hours. But nuclear went off the rails, and now is in deep trouble.
Nuclear plants built 40 or more years ago are reaching a point when they need to be shut down, while new construction has dropped to almost nothing. Less than 5 new plants begin construction every year worldwide, while there are only 50 plants in the process of being built.
Into this debate has stepped Michael Shellenberger, an advocate of nuclear power, who previously had been an environmental activist for decades. In his new book “Apocalypse Never” which has leapt up the bestseller chart, he argues that nuclear is the answer to climate change, while renewables, like solar and wind, are flawed. But his arguments are weak and don’t make much sense.
Shellenberger is correct that nuclear is a reliable producer of carbon-free power. But his discussion leaves out any sense of perspective or the serious problems with nuclear power as a solution to climate change. You wouldn’t know from this book that there are currently a mere 435 nuclear reactors worldwide, 97 of which are in the US, with a combined capacity of 370 gigawatts and producing 10% of world electricity. But there are only 46 new plants under construction, 27 of which are facing significant delays (the average time to build a plant is 10 years). Even China has cut their previous ambitions for 20 new nuclear plants and are currently only actively building 10. Shellenberger doesn’t explain why new reactor starts have fallen from 50 in 1975 to less than 5 currently.
He then compares the amount of power generated by money invested in nuclear and renewables since 1965, and states nuclear has so far come out on top. True, but nuclear became commercially viable in the 1960’s, and renewables only did so in the last 5-10 years. RE costs are plummeting rapidly, completely changing the picture. Lazard (an investment bank) in 2018 found the Levelized Cost of Energy (LCOE) for new wind and solar to be 20% of that for new nuclear, and that nuclear LCOE will rise 23% in the 2020’s while RE continue to plummet. Compared to 1997, by 2018 wind was annually producing 1,258 more Terawatt hours of power (TWh), solar 584, and nuclear only 299. Total nuclear power generation worldwide in 2018 was still 2,563 TWh, but that reflects capacity built before 1997 and still online. These numbers take into account that solar only generates during daylight and wind when wind is blowing, the so-called problem of intermittency which limits renewable energy.
Shellenberger ignores the immediate history of RE. Costs have fallen so much that deployment is exploding, in both rich and poor countries. The first terawatt (a terawatt is 1,000 gigawatts) of RE was installed by 2018, and the second will be in by 2023. By 2030 we will likely have four terawatts, and deployment will continue to accelerate as costs keep falling and storage solutions become cheaper to handle intermittency. Meanwhile, the most optimistic scenario is that nuclear will bring less than 100 gigawatts of net new power online by 2030, a drop in the bucket. Already, China produced in 2018 more from wind alone than nuclear (wind 366 TWh, nukes 277 TWh, solar 177 Twh) and India gets more from wind and solar than nuclear.
One of the benefits of nuclear is unlike solar, it can run 24/7. This is theoretically true, but not so much in practice. France has major reliability problems, with its nuclear fleet only able to operate at 70% of capacity, and plants are often shut down. On an extremely bad day, August 28, 2018, 27 of France’s 58 nuclear plants were offline, and the average French reactor is offline 90 days per year. The American nuclear fleet became extremely well run in the 1980’s, and now can maintain a 90% capacity factor, the best in the world, which allows it to provide 20% of US electricity demand. The marginal cost of electricity from existing nuclear plants is about the same as solar or wind, but that ignores the massive capital costs of new build, which makes new nuclear power simply untenable. The US nuclear fleet has a total capacity of 100 gigawatts, but is on average 39 years old, and starting to age out. Only 8 plants are less than 10 years old. Several plants are only able to stay open due to zero emission credits effectively subsidizing them. Shellenberger simply doesn’t tell you any of this.
Shellenberger also claims that if nuclear is not used then “fossil fuels must be used”. This is simply not true. When California closed San Onofre nuclear power station, it lost 19 TWh/year in generation, but replaced that with 47 TWh from renewables and energy efficiency. Shellenberger lives in California but does not understand California’s energy history. He rightly notes that Californians pay a lot more for electricity, in fact, we pay on average 16 cents per KWh, compared to a national average of 10 cents. But that leaves out the most important part. Up to 1970, US and California per capita electricity consumption was the same and rose in tandem. After 1970, California pursued an aggressive policy of energy efficiency, and while US consumption rose 100%, California has stayed flat. Californians are charged 60% more for power, but use half as much, so as a percent of income, electricity is a smaller burden on people’s budgets in the Golden State. In fact, California households spend less on energy than just about any other state in the nation, despite high electricity and gasoline prices. I’ll take the California approach.
He then moves on to Hollywood and Ralph Nader, and blames them for souring America on nuclear power. But the collapse in new nuclear starts after 1975 was a global phenomenon, not just American. He does not grapple with or explain the immense financial problem of nuclear power.
Shellenberger never does the math on what a nuclear-powered world would mean. How many reactors would need to be built over the next 30 years if we wanted to be zero carbon by 2050? What would that cost? Assuming we would need at least 5,000 gigawatts of global capacity (likely a marked underestimate), we would have to start construction on 250 plants per year. Even at 7 billion dollars per plant, we would need to spend almost two trillion dollars annually. These are completely absurd numbers.
The main impediment to nuclear power is not irrational fear by the public of radiation. The main problem is the massive capital required and the cost of that capital. Shellenberger never bothers to pencil that out. If a company wanted to build a power plant today in the US, it would need access to 7 billion dollars, and would not start generating a return for ten years. If a power plant was able to sell its output 24/7, and could operate 90% of the year, it would sell about 7,900 GWh, at a price of 10 cents per KWh, that would generate about 800 million dollars. If the company secured a 40-year loan at 8% interest, it would pay 500 million dollars every year just to cover the loan. Even if there was a government guaranteed loan with a subsidized 4% rate, the company would still need to come up with 300 million dollars per year. What if there are massive cost overruns? Who would provide this amount of capital for 40 years? Who would take on the bankruptcy risk? If PG&E, Shellenberger’s own local power company, recently emerged from bankruptcy, decided to build five nuclear power plants, who in their right mind would provide 35 billion dollars? Because of the massive capital involved, a nuclear plant needs guaranteed revenue for forty years, which means they can only exist in a highly regulated power market. Usually this means a government guaranteed price and commitment to buy all the power output of the plant for forty years. In a free market, nuclear would have to compete with other power providers, for example solar companies providing power for 2 cents per KWh during daytime. Or wind providers. In a truly free market, nuclear would be the high cost power of last resort, which would destroy its economics as it has such massive capital costs it has to sell its full output 24/7. This is why no deregulated electricity market in the world has any nuclear plants under construction. It’s got nothing to do with irrational fear, which I don’t think figures in Chinese or Indian decision making.
While ignoring the problems with nuclear, Shellenberger spends much of his book attacking renewable energy as a failure. Shellenberger is right that the main drawback with both technologies is intermittency, which is why solar has a capacity factor of only about 25% and wind about 35%. But his economic analyses are deeply flawed. He denounces rooftop solar because the payback is so long for a system that can cost 25,000 dollars, but neglects the obvious solution of financing it. I put solar on my roof in 2016 for 25,000 dollars. My house uses about 12 MWh per year, and I used to pay about 3,000 dollars per year to SCE. I financed my solar system at 6% over 20 years, and I pay less monthly then I did to SCE. My payments are fixed against inflation, and go to zero in 20 years. I did not buy the Powerwall storage system, but that is a gimmick for survivalists, there is no need for storage for residential solar. Net metering solves that issue. Storage is very expensive, but the costs are dropping rapidly, and that will change everything in 10 years, just as utility solar went from an expensive vanity in 2010 to the cheapest power of all in 2020.
Shellenberger misstates the cost of a 100% solar plus storage or wind plus storage grid. Getting that last 10% accounts for much of the cost, because you need a lot of wasteful storage that mostly sits idle. The same numbers would be generated by a hypothetical 100% nuclear grid. In fact, solar (daytime power, more in summer) and wind (more in evenings and winter) complement each other. In addition, wide-scale integration of power and complex demand management can offset much of the intermittency limitations. There will of course need to be a massive increase in storage, but costs are dropping and new technologies are likely to come along. A recent analysis from UC Berkeley shows that the US grid can be 90% carbon free by 2035 at a cost that does not raise electric rates. Using natural gas peakers makes the most sense as they have the lowest capital costs, and natural gas is both cheap and much lower in carbon emissions than coal.
He then talks about the slaughter of birds and insects by windmills. They spin at 10-20 rpm, so not exactly buzzsaws. Still, this obviously happens, but Shellenberger offers no data on who is being killed and how do the numbers compare to other sources of bird and insect deaths (car windshields for example, or cats). What bird species have been driven extinct by windmills? Is it not likely that birds will eventually adopt flight and migration patterns away from these? Siting windmills to minimize these hazards is important. Shellenberger does not mention offshore wind, which is actually far better and more reliable source of wind power, and will likely become the dominant wind energy in the 2030’s. Offshore windmills are unlikely to harm wildlife in a material way or cause bird extinctions.
Shellenberger states that “no amount of technological innovation can solve the fundamental problem with renewables” because they are “unreliable and energy dilute”. I personally could imagine some technology that would solve the problem, but regardless, I don’t know what “energy dilute” means, electricity is electricity and my home seems to run just fine on my solar panels. If he is referring to the amount of surface area needed to generate that electricity, he needs to actually do the math.
He claims that solar panels generate 50 watts per square meter. That’s a little light, modern panels can do 125 watts, but perhaps he is taking into account capacity issues. Let’s use his numbers. Even he concedes that 18,000 square miles would be enough to provide the entire electric power needs of the US. Why is that in any way a prohibitive amount? He also pointed out that an area the size of Alaska has been returned to nature due to improved agriculture, and that is 660,000 square miles. We can’t use 18,000 for solar power? We use more land area to grow corn for ethanol, why don’t we just use that land? Or the 20,000 square miles currently leased out for oil and gas drilling by the US government? He then makes much of the fact that the needed storage would take up another 250 square miles. I find his arguments here completely unpersuasive. He then says that if the US used solar for all its power needs (I assume this means all cars are electric, and so is heating and industrial), it would require 50% of the US surface area. That would be a bit under 1.9 million square miles, and he clearly misstated his point there as that makes no sense. California now legally requires all new homes to have solar roofs, does Shellenberger oppose that? In fact, we could find enough room to power much of America just putting solar on existing homes and buildings and empty spaces like office parking lots.
Shellenberger appears to believe that the size of the physical footprint of a power plant is the most important factor determining its attractiveness. If we developed fusion power, and it used half as much land as nuclear fission, but cost five times as much, would he suggest we have to switch to fusion? I don’t think so, because fission power is not land constrained, so the increased density of hypothetical fusion is practically meaningless. There is plenty of room for solar and wind, they will take up a tiny fraction of land, and far less than humans use for agriculture and grazing. Prices of renewables are plummeting so fast that they will create essentially free power during much of the day. It is not just skittish Americans and Europeans with an irrational fear of nuclear power that is holding the industry back, it is the horrible economics of nuclear power that make it unviable.
Nuclear power is a dying industry at present. Whether new techniques and designs can make it economical in the future, especially in an open competitive power market, seems unlikely in the near future. The only advantage of nuclear is its ability to run 24/7 when well-managed. Whether that is sufficient to create fresh interest in this technology will depend on how well RE and storage and grid management work to keep the lights on in the 2030’s and beyond.
