The US may not be good enough at large infrastructure projects to do it well
We are currently in the midst of protracted interest in a “nuclear renaissance,” including newfound support amongst some environmentalists concerned enough about climate change to bracket fears about nuclear waste and risk and argue for a role for nuclear power. There are four modern reactors under construction in the southern US, which if completed would be a significant step forward after 20+ years of no new reactors coming on line. And government-led expansion of Generation III and III+ reactors has been rapid and relatively inexpensive in South Korea and China.
Despite this, I remain skeptical about the US significantly expanding its nuclear generating capacity as a way to mitigate climate change in the next several decades. Specifically, I think that such an expansion would require a large push of funding and leadership from the federal government that would probably have to go beyond a simple price on carbon, and I think that would be a poor investment based on the US’s recent track record with nuclear power plants and other large, complex infrastructure projects.
There are many other possible reasons to think the US shouldn’t make such a push, and some of them partially influence my assessment. Intergenerational ethical problems top many people’s lists, as politically embattled nuclear waste that needs to be contained for thousands of years is not the kindest inheritance. Fears of catastrophic risk including terrorism and weapons proliferation are also prominent concerns, and are near the top of my list. There are also other worries that don’t sway me as much but are significant parts of the public debate, including nuclear exceptionalism, the idea that nuclear contamination is a unique kind of harm to humans and the environment that cannot be traded off against other costs and risks.
These kinds of concerns are enough to make even the highly climate-motivated reluctant about nuclear power, and I think the final deciding factor is the significant uncertainty surrounding how quickly the US could really build new plants, and at what cost, especially when nuclear cost curves appear to be increasing. In fairness, much of this uncertainty comes from experiences with interminable construction delays in the 1980’s that were often the result of escalating regulations during construction, or public opposition in certain parts of the country. The 2005 Energy Policy Act streamlined many of the most problematic aspects of plant licensing, and the four new reactors under construction are in Georgia and South Carolina where the public is largely supportive of nuclear energy, hopefully paving the way for easier construction.
But even these four reactors are already experiencing significant delays and cost overruns. The two AP1000 units at Plant Vogtle began construction in 2013 and have already been delayed until at least 2019. With capital costs nearing $15 billion for 2.22 gigawatt (GW) of capacity, a basic Levelized Cost of Energy (LCOE) calculation suggests a break even price of around $0.14/kWh. The two AP1000 units at the VC Summer Generating Station began construction shortly before Plant Vogtle, and are also delayed from their original 2017-2018 completion time (2017 for the first unit, 2018 for the second) to 2019-2020. Costs have also escalated, from $9.8 billion to at least $11.2 billion. This yields an LCOE estimate around $0.1/kWh. This might also fit into a larger trend of US struggles with large infrastructure projects, including notably more expensive subway construction costs than other countries, and significantly more difficulty planning high-speed rail.
Of course, much time has passed since our last construction of plants, so delays and high costs aren’t totally surprising. Maybe if we committed to building many more AP1000’s in a row, then costs and construction times would eventually come down and yield relatively dispatchable and inexpensive low carbon electricity. A large entity like the US government could afford to make such an investment, but it doesn’t seem like a good bet to me given the alternatives. First, the size and complexity (both engineering and regulatory) of modern nuclear plants along with the long time scales for licensing and construction make learning-by-doing more difficult than for other low carbon generators. The extreme contrast is solar photovoltaics (PV): many 100 MW solar PV arrays are being rapidly installed in several months or less, and PV cells are being manufactured at a fast pace, creating greater economies of scale and allowing for more incremental advances than the nuclear plants that take years to license and at minimum 4 years to build. Furthermore, such large projects as nuclear reactors are almost certainly more likely to experience significant delays, and this is especially true of plants where regulatory scrutiny of any changes during construction is intense and time consuming. Lastly, nuclear reactors are probably the only low carbon generators that could fall completely out of public favor as the result of one discrete event – an act of terrorism, the use of a weapon, or a significant accident could all lead to irreparable reversals of trust by the public and thus the government. While the chance of this happening in any one year is small, if we imagine making a large push for learning-by-doing that could take several decades it starts to be a considerable risk.
This isn’t to say that there will be no new nuclear reactors installed in the US in the future. It’s quite likely that there will be at least several more, and it’s possible that costs could come down significantly after this first new wave of reactors is built and spawn a large, spontaneous build-out. There seems to be a strong possibility that China will expand its nuclear fleet, likely benefiting from a strong centralized government and a track record of timely construction. But it has been a long time since the US has built reactors economically, and relative to other countries we might have a harder time executing large, high profile infrastructure projects, especially if they draw significant public interest and possible litigation. This leads me to believe significant government support would be needed to make nuclear expansion a reality, and that it would not be a wise choice even viewed strictly as a carbon-reduction strategy. Momentum matters when tackling a contentious issue like climate change, and the US might be better off putting its effort behind technologies with cost curves that are more obviously declining, and that can be built in a series of smaller victories rather than large, one-GW steps that could be contentious or frequently delayed.
 See LCOE calculation in Literature Cited section.
 See LCOE calculation in Literature Cited section
 See Lepska (2011) for comparison of per-km costs of subway construction in different cities, and Dayen (2015) for a brief review of the sources of delays and opposition to high speed rail in California.
Daniel Thorpe is a PhD candidate at Harvard School of Engineering & Applied Sciences.
References of the Article here.