How Long Does It Take To Build a Nuclear Reactor?
Let’s take a look at the data.
That the EU has moved to classify investment in natural gas and nuclear energy as green has caused an outcry across European civil society, albeit to different people for different reasons. While some praise gas as a lower-carbon transition fuel on the way towards 100% renewable energy, others point to France’s low energy-related emissions and promote more investment in the nuclear sector. Now, the risk of gas shortages due to the Russia’s invasion of Ukraine has rekindled the debate around the role of nuclear power in the energy transition.
Nuclear energy is a hot topic
Data from the IPCC (2014) shows that nuclear energy is one of the least carbon intensive energy sources we have. At about 12 gCO2eq/kWh, only wind power emits less greenhouse gases per kilowatt-hour.
Despite some high-profile disasters, nuclear power is also one of the safest sources of energy we have. According to estimates compiled by OurWorldInData, each terawatt-hour of nuclear energy is responsible for about 0.03 deaths. For perspective, compare that to the 24.6 deaths/TWh from coal (mostly due to air pollution).
Among the numerous problems with nuclear power (storage of nuclear waste, centralization of energy production, nuclear weapons proliferation, national security concerns, price, upfront capital cost, fuel availability, etc.), a factor that has the most practical impact on its success for the transition away from fossil fuels is how fast we can build nuclear power plants. Construction time is one of the main problems of nuclear power limiting its usefulness in addressing climate change— a problem we have to solve right now.
The argument goes that nuclear power plants would take too long to build to be of any use now. So, how long does it take to build a nuclear reactor?
Reactor construction time
The International Atomic Energy Agency (IAEA) has compiled a neat overview of all nuclear reactors — those in operation, in planning, under construction, and even those permanently shut down.
Each dot on the chart corresponds to one nuclear reactor. A nuclear power plant typically consists of several of these units. The color of each dot indicates when construction started. We tended to build smaller (often research) reactors in the early days of the nuclear industry, and then moved to larger units around 1 GW as the industry matured.
What we are analyzing in the diagram above is the time from the start of the construction (i.e., the first pouring of concrete) until the reactor is connected to the power grid for the first time. A similar analysis has been done by Carajilescov et al. in Energy Policy.
We then average this construction time over all operational and shut down reactors. The average construction time is almost exactly seven years. This figure comes with a large spread, however: while the shortest construction time was just under two years, the longest took almost 43 years to complete! The standard deviation is about four years, meaning that about 68% of all reactors are completed between three and eleven years.
How does a reactor’s capacity factor in? Surely a larger reactor like China’s Taishan units (1660 MW each) will take longer to build than a research reactor like Germany’s first test reactor in Kahl (150 MW) or the world’s smallest reactor still in operation, Russia’s Bilibino reactor (11 MW).
Plotting reactor construction time vs. electrical net capacity reveals an interesting relationship. While the construction time does increase with capacity, it does so only to a small degree. One can speculate what that means for small modular reactors. Based on a best-fit linear regression, we can estimate the average reactor construction time to be:
average construction time = 5.5 years + (capacity × 2.0 years/GW)
The most common reactor capacity — as seen in the histogram at the top of the chart — is about 1 GW (=1000 MW). The linear regression tells us that the average construction time of such a reactor would be about 7.5 years.
So there we have it! A usual reactor takes about three-quarters of a decade to build. If we started right now in 2022, we could have shiny new reactors by 2030.
However, the linear regression obviously fits only very badly. Instead of claiming a linear relationship between capacity and construction time, it might be more correct to say that there is no meaningful relationship in the first place. We might also want to exclude some of the outliers, some of which have had their construction paused for a long time. Excluding the outliers does not improve the fit in a meaningful way, however, and does not change the average much.
We might also want to look at the relationship between construction start and construction time. Surely we’ve gotten better at building reactors over time?
In this chart, we now plot the construction time vs. the construction start. The colors correspond to the installed capacity (lighter = smaller reactor). Outliers, i.e., the reactors with really long construction times above 20 years, are excluded from this chart. This affects only 8 out of 640 reactors that have ever fed power to the grid.
We can see that in the early phases of the nuclear industry, mainly smaller reactors were built, with an overall shorter construction time. If the construction time would have improved over time, i.e., with more experience, we’d expect the dots to gradually assemble on a decreasing average. This doesn’t seem to be the case. From 1965 onwards, the reactor construction time and reactor size do not seem to have any real relationship with the construction start. At best the construction time went slightly up as time goes on.
Caveats
From these plots, one could draw the conclusion that if we wanted to, we could easily build new reactors in a relatively short time. There are a number of caveats to that statement, though.
What we looked at is the construction time only, from the first pouring of concrete until the first grid connection. It does not take into account the excessive planning and approval phase prior to construction. According to Shykinov et al., the activities towards construction and comissioning typically add 11-12 years on top of the construction time. Berthélemy et al. highlight the microeconomic disadvantage this comparatively long lead time causes:
[…] one might prefer to build a CCGT gas plant that can be planned and built in 2 years and be willing to face the fossil fuel and carbon price risk, instead of waiting more than 7 years (in the best case scenario) to start recovering their investment.
Furthermore, we are only considering the time it takes until the first grid connection. In general, this does not correspond necessarily to the time the reactor operates commercially at full capacity.
If we wanted to start construction of new nuclear reactors now, we would also have to take into account the preliminary delay and the changing energy market situation by the time of completion. As the cost (levelized cost of electricity) continues to drop rapidly for solar power, the economic case for new nuclear power plants becomes more difficult.
With proper maintenance, nuclear reactors can run for a long time — India’s Tarapur reactor blocks are running for about 52 years already, for instance. This mitigates the disadvantages the construction time brings with it somewhat, but does not alleviate the difficulty in achieving the “explosive” nuclear reactor construction spree necessary to meet the 1.5°C goal with nuclear power alone.
Whether nuclear power can be considered renewable is another question. While sufficiently enriched uranium is somewhat limited, fast breeder reactors (like France’s Superphénix) in conjunction with reprocessing facilities (like France’s La Hague site) can overcome this limitation, but bring with it increased risk of nuclear weapons proliferation.
According to the IPCC report mentioned earlier, nuclear power will play a role in the energy transition. How big of a role it is capable of taking on, however, remains to be seen. ⬢
Methods
Tools: The infographic was created using Microsoft Excel and OriginLab for basic plotting and statistical analysis. I used Adobe Illustrator to pretty it up and add additional information. You are welcome to check out my other work on Behance.
Data: Reactor construction time and installed capacity is taken from the IAEA’s Nuclear Power Reactors in the World 2021 report. They provide helpful downloads for the raw data as well (“Supplementary Data”). The information of the individually highlighted reactors is taken from the corresponding Wikipedia pages.
How Long Does It Take To Build a Nuclear Reactor? was originally published in Climate Conscious on Medium. If you enjoyed reading this article, you might be interested in subscribing to Remember Your Humanity for more content.
It would be interesting to see the data split by state at the time of construction, since I would expect that regulation is the strongest factor in determining the construction time.