Fusion and Your Future Electric Bill

Nuclear Fusion Part 4: Cost Assessment

Key Takeaways:

• Commercial fusion power is 10–15 years away. I expect my toaster to run on fusion by 2035.
• Though the cost of solar and other electricity sources will continue to fall, fusion is likely to be economical (a napkin-level analysis suggests a levelized cost of electricity of $0.06–0.11/kWh).
• The cost of fusion power is highly dependent on factors such as the size of the power plant, the cost of converting steam into electricity, and the discount rate (cost of capital).

In this post, we’ll look at when our toasters could run on fusion power, and how expensive this will be. There is much we still don’t know. However, by stating our assumptions and walking through what the future might look like, we can start to gauge what is achievable in reality vs what is science fiction.

When Will We Get Commercial Fusion Power?

My crystal ball says 10-15 years. Let’s explore why.

The first step to fusion power is getting to net energy gain, where the energy produced by fusion is greater than the energy needed to heat the plasma. In previous posts on magnetic confinement fusion (magnets) and inertial confinement fusion (lasers), we discussed some of the challenges to achieving net energy gain, or ignition. Because of these challenges, advanced facilities are needed: giant lasers, magnets, materials that can hold a mini-sun, and all the controls and diagnostics to make them work. These facilities may cost hundreds of millions or billions of dollars, depending on the design (more on this below), and it will not happen overnight.

Still, the race to achieve net energy gain includes a growing number of players who have made impressive strides. Historically, large-scale government research projects like ITER have driven progress in fusion. This may shift in the next decade, as more independent companies raise hundreds of millions of dollars while leveraging the lessons learned from extensive public research efforts. Although the fusion sector lacks an explicit late-stage public-private program, the convergence of public and private capital is similar to NASA’s Commercial Orbital Transportation Services (COTS) program, which supported the development of SpaceX.

The largest government-led fusion effort, ITER (the international experimental tokamak project described in Part 3), is currently scheduled to begin its first experiments in 2025. Deuterium-tritium (DT) experiments that could achieve net energy gain will start after 2035.[1] However, ITER doesn’t include the back half of the power plant, which turns the heat from fusion into electricity. ITER will conduct experiments to optimize the design for DEMO, a larger tokamak fusion facility that is slated to generate electricity in 2050. [2] Fusion energy through the ITER program is currently at least 30 years away.

The timelines targeted by private companies and startups are much faster. Some have publicly announced plans to achieve net energy gain in 2025 or before. MIT-backed Commonwealth Fusion (which has designed a smaller, more compact tokamak) recently raised an additional $115mm to complete their concept and achieve net energy gain in 2025. The UK’s Tokamak Energy also targets net energy gain midway through the decade, and first electricity to the grid in 2030. Several inertial confinement (laser) startups including Marvel Fusion, HB11 Energy, and Innoven Energy have also emerged to leverage years of research at labs around the world.

Crossover concepts (using a combination of magnets and some source of inertia to confine the plasma) offer even more shots on goal. [3] These fusion companies include relative veterans TAE Technologies (founded 1998) and General Fusion (2002), established startups like Helion (2013), and newcomers such as those accelerated by ARPA-E’s ALPHA fusion program. Some of these new companies aim to reach net energy gain before 2025 through smaller, more compact fusion devices.

Within the next 5–9 years, I believe one of these teams will demonstrate net energy gain. From there it will likely take an additional 5-10 years from net gain to getting electricity on the grid, assuming:

• The world funds enough fusion designs that one of them works.
• Regulators are able to play ball and create pathways for fusion companies to obtain operating permits.
• No more pandemics / giant meteors / other disasters push schedules back.
• The electricity produced is cost competitive.

The relevant math here:
5 years (to net energy gain) + 5-10 years (to complete the power plant) = 10-15 years

I am optimistic that we will see the world’s first fusion power plant by 2035.

What Will Energy Cost in 2035?

When fusion power breaks onto the scene in 2035, will we want to pay for it? Or will prices for other sources of electricity continue to fall, making it difficult for fusion to compete? The question of what energy prices will be in the future is not an easy question — entire organizations and thousands of experts around the world are working on predicting the future of electricity markets.

The easy answer is that somewhere electricity prices will likely be high enough to support fusion power. In 2019, the average retail price of electricity in the US was $0.11 per kilowatt hour (kWh). In Hawaii, the average price was $0.29/kWh; in Louisiana, it was below $0.08/kWh). Electricity retail prices tend to be higher in western Europe and Japan — residential prices of over $0.25/kWh were reported in Germany, Denmark, Belgium, Italy and Ireland. The key takeaway here is the variability of electricity markets from place to place.

Today in the US, we have about 1,100 GW (gigawatts, or billions of watts) of electricity generating capacity according to the US Energy Information Agency (EIA). The EIA assumes that US electricity demand will grow at roughly 1% per year on average through 2050. The additional capacity we need from 2030 to 2050 to cover this and replace power plants that retire is roughly 20 gigawatts (GW) a year. For context, 20 GW is about 25 natural gas plants [4] or 20 nuclear power plants.

Source: U.S. Energy Information Administration, Annual Energy Outlook 2020 (graph)

The EIA currently predicts that most of this growth will come from solar and (to some extent) natural gas, based on an assessment of what will be most cost effective. This assessment uses a metric called the “Levelized Cost of Electricity” or LCOE to compare options for electricity production. [5] The LCOE is basically the sum of all the money it takes to build the power plant and operate it for its entire life, divided by the amount of electricity it produces over those years. For those who like math:

I = capital costs, M = yearly operations and maintenance costs, F = fuel costs, and E is the amount of electricity generated. Careful! There’s a sneaky discount rate (r) in there.

Based on the predictions from the US Energy Information Agency (EIA), the cheapest form of energy in 2050 will be solar with an LCOE of $0.026/kWh - it will be dispatched whenever possible. Onshore wind energy will cost on average $0.035/kWh and electricity from natural gas will cost about $0.045/kWh. [5] It’s important to note that the LCOE is NOT the retail price on your electricity bill. The LCOE does not include transmission and distribution costs, taxes, and marketing — it’s just the cost of making the electricity.

To convert dollars per megawatt hour ($/MWh) to $/kWh, divide the LCOE in megawatt hours by 1000. Source: U.S. Energy Information Administration, Annual Energy Outlook 2020 (graph)

How Will Fusion Power Compare?

Since no fusion power plants have been built yet, we have two options for estimating what fusion power plants might cost in the future. First, we can review the few public engineering cost estimates that are available for fusion concepts. Where information is not available, we can look at the cost of projects with a similar scale and level of complexity (e.g. fission power plants), and extrapolate.

To demonstrate what this might look like, let’s walk through a “back-of-the-envelope” estimate of the levelized cost of electricity (LCOE) for a fusion power plant. Warning: to complete this thought experiment, we will make some significant assumptions, and do things that the Advanced Research Projects Agency — Energy (ARPA-E) specifically tells us not to do. [6]

In 2017, the US ARPA-E, Bechtel and technology providers worked together to estimate the potential cost of a small fusion power plant based on four technologies developed through the ALPHA program. Each plant was designed to generate 150 MW of electricity. The estimated total overnight cost (TOC) for the power plants averaged $1.3B (in 2016 USD dollars), with estimates for each technology ranging from $0.7B to $1.9B. These costs assume that each plant is the 10th plant built, i.e. that most of the kinks have already been worked out.

Say Company A successfully demonstrates fusion and can build a power plant that produces 150 MW of electricity for $0.7B. We don’t know how much it will cost to operate the fusion plant, but let’s assume that operating costs are roughly 40% of the LCOE, as for nuclear power plants, and that it will operate for 30 years. [7] The LCOE for this fusion plant would be $0.11/kWh. Considering EIA’s predictions of about $0.03–0.045/kWh for solar, wind and natural gas in 2040, it is unlikely that fusion will be the absolute cheapest source of electricity. Still, fusion may be economically attractive for small power plants in remote areas with low sunlight or access to other fuels (remember that Hawaii’s retail electricity price is almost 3X the national average).

The simplest path to reducing the LCOE is increasing the size of the power plant to access economies of scale. Using engineering rules of thumb, Company A’s technology scaled to 1000 MW (the size of a typical nuclear power plant) would cost roughly $2.2 B. [8] Now the predicted LCOE is $0.052/kWh for the same technology. Extending the life of this plant to 40 years would further reduce the LCOE by spreading out the capital cost over a longer period.

The breakdown of predicted equipment costs are also interesting. In the four small fusion plants considered, the cost of the core fusion equipment — the chamber, magnets, lasers, fuel injection, and converting fusion products to heat — makes up only 28–45% of the total capital cost. The steam turbine and electrical equipment was a significant 21–34% of these costs. Perhaps finding ways to convert or repurpose retired power plants could reduce these costs. The cost of structures and site facilities was also a large fraction at 21–28%.

Cooling towers with large steam plumes show that all 4 nuclear reactor units at the Cattenom Nuclear Plant are running. Fusion power plants would likely have similar steam turbines and cooling towers. (Image Source)

Converting the fusion products directly into electricity without first making steam (to drive a turbine) would result in major cost savings. This is a big reason that some fusion companies hope to use proton-boron fuel, even though it is harder to ignite. The proton-boron reaction produces atoms with an electrical charge — this means that some fraction of the energy can be converted into electricity directly.

For the sake of simplicity we have focused here on a recent and relatively detailed fusion cost estimate that is publicly available, the 2017 ARPA-E/Bechtel study. Techno-economic assessments of fusion technologies are currently scarce [9], and will be increasingly important as they approach commercialization.

Final Thoughts

For fusion to be widely adopted as a clean, reliable and safe source of electricity, we need to start thinking of fusion not just in terms of awesome science, but also in terms of power plants. 2035 is not that far away.

To support this, ITER and other government programs have focused more attention and resources on the engineering challenges associated with fusion. These include fuel production and handling, and materials for the “first wall” between the plasma and the energy conversion equipment. The ARPA-E fusion program recently announced up to $30mm in funding through its GAMOW program targeting these areas. More funding will absolutely be needed to carry promising fusion technologies through to commercialization. However, the ability to unlock billions of dollars of low-carbon, reliable energy at economical price points for hundreds of years into the future makes investment in this space worth it.

Nuclear engineers at work (own work)

I gratefully acknowledge the help and input of fusion scientists and engineers and some brilliant people at ARPA-E who informed this blog post. As always any mistakes are my own, and I would welcome the opportunity to correct them.

The first post in this series was a high-level overview of the fusion space. In the second post, we explored inertial confinement fusion technology. The third post looked at magnetic confinement fusion. In this final post, we considered what a fusion plant might cost, and how it might fit into the future energy landscape.

Notes

1. While the ITER tokamak will begin running basic deuterium fusion experiments in 2025, handling radioactive tritium requires specialized equipment and materials. The target for completing these upgrades is 2035.
2. The ITER facility is designed to perform a range of experiments and measurements. The DEMO design will be larger but simpler, and will include equipment to actually convert the heat from DT fusion into electricity for the local electric grid. Another important task for ITER is to produce enough tritium for commercial facilities, and specifically for DEMO to use while starting up. Once a fusion plant is running on tritium, it can produce more through “breeder” reactions with other materials.
3. The Fusion Industry Association lists even more active fusion companies for those interested.
4. The average size of natural gas plants in the US is ~800 MW. Source: EIA
5. Source: EIA “Today in Energy”, March 6, 2020.
6. The ARPA-E report clearly states:
   “ARPA-E and the technology providers understand that the report does not contain sufficient accuracy or detail to be meaningful in connection with any securities offering or other financing effort and due to the status of the technology development, the uncertainties as to time horizon in which any of these technologies could be commercially deployed, and the limited scope of the review, this report is not intended to be relied upon by any third party in making investment decisions.”
7. All designs generated roughly 500 MW of heat, and converted this heat to electricity with 40% efficiency. The distribution of capital costs and operating costs for nuclear power plants (and other electricity generation) was taken from Figure 1 in Hirth and Steckel, 2016. We’ve also assumed a discount rate (r) of 10%.
8. This is where we break the rules a little. Let’s apply the “6/10ths” rule to estimate the costs of engineering equipment for a larger 1000 MW facility, starting from our 150 MW power plant that costs $700mm. We project a capital cost of $2.2B at the larger scale. Starting from the average cost of $1.3B in the ARPA-E report gives a cost of $3.7B for a 1000 MW power plant.
   In an unconstrained design space, this rule determines the power plant size proposed by fusion companies to be profitable.
9. Other studies with estimates for fusion power costs:
   - Electric Power Research Institute (EPRI), Technical Report 1025636, Oct 2012.
   - Entler et al, “Approximation of the economy of fusion energy”, Energy 2018. (A techno-economic assessment based on a design for DEMO)

Prime Movers Lab invests in breakthrough scientific startups founded by Prime Movers, the inventors who transform billions of lives. We invest in seed-stage companies reinventing energy, transportation, infrastructure, manufacturing, human augmentation and computing.

Sign up here if you are not already subscribed to our blog.