Had the Titanic set sail today, it might not have met the same tragic end. But the disappearance of icebergs from our oceans is anything but good news. Modern-day endeavors to slow the pace of global warming and avert a climate disaster are disturbingly similar to the efforts of people over a century ago who believed the ocean liner was unsinkable. We know we’re getting close to the point of no return, and that we’ll need to achieve global carbon neutrality by 2050 in order to keep warming below 1.5°C above pre-industrial levels. Yet according to an International Energy Agency (IEA) report issued ahead of the COP26 conference, pledges made to date amount to just one-fifth of what is required to limit the global temperature rise to 1.5°C by 2030.
In short, the world isn’t doing nearly enough. The challenge for humanity is to reduce, and ultimately halt, atmospheric CO2 emissions, which last year amounted to a staggering 36 gigatons. In its report, the IEA called for “a massive additional push for clean electrification that requires a doubling of solar and wind deployment relative to the APS [the IEA’s Announced Policies Scenario]; a major expansion of other low-emissions generation, including the use of nuclear power where acceptable; a huge build-out of electricity infrastructure and all forms of system flexibility, including from hydropower; a rapid phase out of coal; and a drive to expand electricity use for transport and heating.”
- Proven technology and disruptive ideas
- Looking for safety 20,000 years ahead
- Meet CROCUS, EPFL’s zero-power nuclear reactor
- In search of the ultimate energy source
- “Nuclear fusion can provide clean, safe and stable baseload power for the future of humanity”
- Towards a multifarious energy mix
One word in this paragraph stands out: “nuclear.” Some 36 years on from Chernobyl and 11 years after Fukushima, atomic energy is back on the agenda — this time as part of the solution to decarbonizing global energy production. As a constant, abundant, zero-emissions source of power, nuclear certainly sounds appealing. Yet fears about serious accidents and the headache of disposing of radioactive waste remain unresolved in many countries, undermining its acceptability.
Interestingly, in early February, the EU added nuclear energy and natural gas to its taxonomy of sustainable activities, making them eligible for green investments. This move, which followed pressure from France and Germany, drew wide condemnation — and not just from environmentalists and anti-nuclear activists. Nuclear power remains a sensitive issue because it cannot be framed in technical terms alone. “Modern-day risk governance principles were, in part, shaped by nuclear risk assessment,” explains Marie-Valentine Florin, executive director of EPFL’s International Risk Governance Center (IRGC). “That’s because technical reassurances about nuclear safety and security aren’t enough to put most people’s minds at ease.” According to Florin, four persistent obstacles block nuclear’s path to acceptability: fear of accidents; concerns about nuclear proliferation and security; long-term waste management issues; and the risk of dependence on uranium-producing countries.
“The International Atomic Energy Agency’s pillars of nuclear safety and security are absolutely crucial, and will also help allay people’s fears,” says Florin. “We’re seeing how important that is today, with the anxiety associated with the Russian army’s seizure of the Zaporizhzhya and Chernobyl power plants in Ukraine.” The recommendations state in particular that: “Operating staff must be able to fulfil their safety and security duties and have the capacity to make decisions free of undue pressure.” Florin further points to the risks associated with the dual use of nuclear power. “These risks will never go away, since nuclear technology is very strategic from a geopolitical standpoint. But they can be mitigated by enforcing strict engineering standards and governance principles, such as a low level of corruption, political stability, government effectiveness and regulatory competence.”
Modern-day risk governance principles were, in part, shaped by nuclear risk assessment”
Decabornizing the energy mix
“A country’s energy mix is very much a political decision,” adds Florin. “Willingness and resource considerations also play a big part.” Andreas Pautz, who heads EPFL’s Laboratory for Reactor Physics and Systems Behaviour, makes a similar observation: “These kinds of discussions are specific to each country. There’s almost a religious element to it. None of this is helpful when our climate is changing so dramatically.” In Europe alone, divisions run deep. In the wake of the Fukushima disaster, Germany pledged to phase out nuclear power within a decade. The country will shut its remaining three plants this year, even if doing so means it could fall short on its climate pledges. Austria, Denmark, Luxembourg and Spain are in the same camp, having stated their opposition to including nuclear power in Europe’s future energy mix. Switzerland, meanwhile, has voted against building any new nuclear power plants. “It was a highly emotional decision, but not an irrational once since there is some logic to it,” says Florin.
Advocates of nuclear power see it as part of the path to a low-carbon future. France currently derives 70% of its power from nuclear energy. Other pro-nuclear European nations include Slovakia (53%), Hungary (48%), Bulgaria (40%), Slovenia (38%), the Czech Republic (37%), Finland (34%), Poland and Romania (both 20%). Plans are in the works for new plants in France, the UK, Argentina and the US, while construction projects are already under way in places like China, South Korea, Russia, Turkey and the United Arab Emirates — countries where there’s hardly any debate about nuclear facilities. “In these countries, a new plant can be up and running within five or six years — on time and on budget,” says Pautz.
A country’s energy mix is very much a political decision”
Comparing life cycles
Political will and safety and security concerns aside, opponents point to cost as another downside of nuclear power — not just in monetary terms, but also in terms of CO2 emissions from construction, decommissioning and mining. “The main problem is that nuclear will struggle to remain competitive as the cost of renewables falls further,” says Marc Vielle, a scientist at EPFL’s Laboratory of Environmental and Urban Economics. Pautz, meanwhile, takes a somewhat different view: “It’s all well and good taking about prices, but it’s a question of perspective. Power generation costs are only part of the story. If you want an accurate picture of the real cost of energy systems, you also have to factor in expenses related to feeding renewable energy into the power grid, building storage facilities to iron out seasonal fluctuations in production, and producing the hefty amounts of materials needed to build wind and solar farms — not to mention the issues surrounding where those materials are sourced.”
“The only way to tell if the benefits of nuclear outweigh its costs, or vice versa, is to compare nuclear power plants against alternatives across their life cycle — and to take location into account,” explains Florin. “The results of these life cycle assessments vary considerably, especially when geography comes into play.”
For all its appeal, the future of nuclear is “far from a bed of roses,” according to Vielle. Pautz strikes an equally cautious tone, albeit for different reasons: “I wouldn’t say that nuclear power is the answer, but I don’t see how we can do without it. With demand for electricity set to double over the next 30 years, we’ll need a stable, zero-carbon power source that can account for at least 20% of the global energy mix, alongside wind and solar.”
The IEA has modeled three scenarios to estimate future energy trends, with each model based on different assumptions about the trajectory of CO2 emissions between now and 2050. The most pessimistic of these is the Stated Policies Scenario (STEPS), which reflects current policy settings. The second is the Announced Policies Scenario (APS), which assumes that all existing commitments will be met. The third — and most optimistic — is the Net Zero Emissions by 2050 Scenario (NZE). Under the first two scenarios, the share of nuclear power in the global energy mix will fall from its current level of 10% to 8% in 2050, although output will increase in absolute terms. Yet in the NZE scenario, nuclear energy will account for 15% of the energy mix, because the model assumes some degree of heat and power cogeneration. “We’ll need to follow the NZE scenario if we’re going to keep warming below 1.5°C,” says Vielle. “But the models also show that we can cut emissions without relying on nuclear power. After all, unlike fossil fuels, nuclear has always been a small part of a much bigger picture globally. Many countries don’t use it at all.” ■
The main problem is that nuclear will struggle to remain competitive as the cost of renewables falls further”
Today’s nuclear energy landscape largely reflects choices made by governments over 50 years ago. According to the latest figures from the International Atomic Energy Agency (IAEA), 439 reactors are currently operating in 32 countries. Roughly 25% of them are in North America (the US and Canada), another 25% in Southeast Asia (China, Japan and South Korea), around 20% in Western Europe, and a slightly smaller percentage in Central and Eastern Europe (Russia and Ukraine). The remainder are mainly in India and the Middle East. The International Energy Agency puts the amount of power generated by the world’s nuclear plants at 2,692 TWh (in 2020), or some 10% of total power demand (26,762 TWh).
Many of these plants are reaching middle age, as over half were constructed more than 30 years ago. The majority of them have generation II reactors (built in the 1970s and 1980s) or generation III reactors (built in the 1990s, with enhanced safety and security measures). In terms of reactor type, 70% are pressurized water reactors (PWRs), 14% are boiling water reactors (BWRs), and 11% are pressurized heavy-water reactors (PHWRs).
Andreas Pautz heads EPFL’s Laboratory for Reactor Physics and Systems Behaviour© Gilles Gonin
Nuclear power plants generate electricity in pretty much the same way as coal- and gas-fired plants. They produce large amounts of steam, which is fed into a turbine and causes the turbine blades to rotate. This activates an alternator that generates an electric current. However, in nuclear plants, the heat used to create the steam comes not from fossil-fuel combustion but rather from the fission — or splitting — of atomic nuclei. The fuel for nuclear power plants is fissile material, or material consisting of atoms whose nuclei can be split apart by neutrons, thus releasing large amounts of energy. The main fissile material used today is uranium 235. Subjecting just one gram of uranium 235 to fission can release just as much energy as burning a metric ton of oil.
During fission, some of the neutrons that are released go on to impact other nuclei and induce more fission, resulting in a chain reaction. This chain reaction is carried out inside a nuclear reactor core, albeit in a controlled manner so as to maintain a constant pace of fission. Uranium used as a fuel can last for three to five years; once spent, it becomes a highly radioactive material that must be stored under highly regulated conditions. Some types of nuclear waste can remain radioactive for over 200,000 years. For now, only Finland and Sweden allow deep geological repositories of nuclear waste. Another promising option is recycling waste into new fuel for certain types of nuclear power plants, but this technology has yet to be demonstrated.
Ninety-five percent of nuclear reactors today are water-cooled. In pressurized water reactors, the water in the primary coolant loop does not come into direct contact with the turbine; instead, it serves as a neutron moderator and heat exchange fluid. It’s kept under high pressure so that it remains in a liquid state even at 300°C. In boiling water reactors, the steam is produced inside the reactor and fed directly into the turbine to generate electricity.
New kids on the block
Some reactors use “heavy” water, in which 99% of the hydrogen atoms have been replaced by deuterium, a hydrogen isotope. This heavy water serves as a neutron moderator, improving the balance between neutrons created and lost and enabling the reactor to run on non-enriched fuel. Another type of reactor is the gas-cooled reactor (GCR); these are used in the UK, where they make up 3% of the country’s total nuclear reactors but will soon be decommissioned.
Fifty-two reactors are currently being built around the world. Over half of them are in Asia (China, India, South Korea and Japan), 12 in Central and Eastern Europe, and 4 in Western Europe. The IAEA reports that together these reactors will have a total capacity of 55 GW. Of the reactors under construction, 44 are PWRs, because plant operators feel more comfortable using proven technology for safety reasons.
“After Fukushima, safety became the top priority in nuclear research,” says Andreas Pautz, the head of EPFL’s Laboratory for Reactor Physics and Systems Behaviour. “Generation III reactors employ the same technology as the previous generation, but have enhanced security features — such as to protect against earthquakes and floods — and a potentially longer life. Now researchers are looking at new technologies for even more advanced nuclear reactors.”
These new technologies are collectively referred to as “generation IV.” They are based on six reactor concepts that differ substantially from PWRs and BWRs in order to meet the following goals: to provide greater safety and reliability, so that people don’t have to be evacuated if there’s a major accident; to enhance sustainability by consuming fewer resources and generating less end waste; to improve cost competitiveness relative to other types of energy; and to better protect against attacks and the risk of proliferation.
After Fukushima, safety became the top priority in nuclear research”
Leveraging the potential of cogeneration
Many of the concepts currently being explored date back to the beginning of the atomic age. Now researchers are taking another look. “People came up with some interesting ideas,” says Pautz. “For instance, one design entails cooling reactors with gas or sodium. These reactors could operate at much higher temperatures than the ones we currently use. Part of the problem is that nuclear plants have always been designed for the sole purpose of power generation, and the excess heat is evacuated through a cooling tower. What I’d like to see are plants that generate both heat and power, with the heat being used to produce hydrogen. This would be a much more efficient way of producing hydrogen than standard electrolysis. The idea with generation IV reactors is to combine heat and power generation in what’s known as cogeneration. That would make nuclear technology a really powerful tool and considerably expand the range of applications.”
Molten salt reactors (MSRs) use hot liquid forms of fluoride and chloride salt at low pressure as the primary coolant. They operate at higher temperatures than conventional nuclear reactors, making them considerably more efficient at generating power. What’s more, they’re designed to work without a solid fuel, thereby eliminating the challenges associated with producing those fuels. MSRs can be adapted to different types of nuclear fuel cycles, such as uranium-plutonium and thorium-uranium, meaning they can draw from a broader fuel supply. And they can be designed to “burn” nuclear waste or to operate as breeder reactors. The large quantities of heat produced by MSRs can be deployed for power generation or other industrial processes needing that kind of energy.
Engineers are also looking into fast-neutron reactors (FNRs) with a closed-loop fuel cycle, an idea that was initially hatched in the 1960s. These reactors are designed to maximize the combustion potential of uranium and especially uranium 238. They employ fast neutrons and operate without a neutron moderator, and they are cooled using a liquid metal such as sodium, lead or lead-bismuth. So far, only sodium-cooled FNRs have been used in industrial applications. These reactors operate at or near atmospheric pressure and have passive safety features. Because the low pressure reduces the chance of a coolant tube leaking or breaking in the event of an accident, FNRs are considered safer than other types of reactors.
One final type of new reactor is the high-temperature gas-cooled reactor (HTGR). These reactors have helium as the primary coolant. They offer excellent fuel conversion rates and can produce heat at 700°—1,000°C, paving the way to the cogeneration of heat, power and hydrogen. HTGRs are inherently safe and don’t require an active safety system. Even in the worst of cases, the reactor core would never reach melting temperatures. HTGRs could be constructed without a containment building.
The idea with generation IV reactors is to combine heat and power generation in what’s known as cogeneration. That would make nuclear technology a really powerful tool and considerably expand the range of applications.”
Small reactor kits
Some countries like China have massive power requirements that can be met only by building super-sized plants. But increasing attention is being paid to the idea of exploiting nuclear energy on a smaller scale through what are known as small modular reactors (SMRs). These are mini-power plants with a capacity of up to 300 MW. They offer greater flexibility since they’re quicker to set up and less capital-intensive. SMRs could be installed on ships, for example, or as part of hybrid power-generation systems combining nuclear with other types of energy, particularly renewable ones. “The drawback to SMRs is they have a priori higher construction cost per installed kW,” says Pautz. “The challenge will be to make them cheaper than regular plants.”
The next generation of nuclear reactors will likely draw from one or more of the concepts described above. They will be equipped with intrinsic and passive safety systems to mitigate risks in case of a malfunction. And they will be cogeneration facilities — producing both power and heat — for maximum efficiency. One new avenue for building nuclear plants is to prefabricate individual modules and then assemble them on site, based on the desired power generation capacity. That would save time and allow for economies of scale.
This could be a particularly attractive option for remote regions or regions with specific power requirements, such as for water desalinization. Russia is one example: thanks to two 35-MW PWRs located off the coast of Pevek, a port town in eastern Siberia, the town’s 4,500 residents can now take a “nuclear shower.” The modular PWRs were installed on the Akademik Lomonosov in May 2020 and supply the town with electricity, heating and hot water — crucial for this region north of the Arctic Circle where temperatures stay below zero for eight months a year.
From Rolls-Royce to Bill Gates
SMRs are in the advanced stages of construction in several countries (including Russia, China and Argentina), and around a dozen more projects are in the works in the US, China, Russia, Canada, South Korea and the UK. Governments and private-sector investors are waking up to the technology’s potential. In the UK, for example, Rolls-Royce hopes to build up to 16 SMRs by the end of the decade.
Many US companies have their eyes set on SMRs, too. In Wyoming — part of the country’s coal belt — Bill Gates and Warren Buffet have invested in a new sodium-cooled FNR pilot plant with a molten-salt energy storage system. The plant should be completed in 2028 and will have a capacity of 345 MW. “By combining these technologies, the plant will be able to generate more electricity with a relatively low-power reactor. And it will function like a thermal solar farm, in that previously stored energy will be released during peak load times,” says Pautz. “I think this approach could be an effective supplement to wind and solar power. If Bill Gates succeeds in building the pilot plant and proving it can operate safely, that could spark a major shift in the industry.”
The International Risk Governance Center (IRGC) — a cross-disciplinary center based at EPFL — issued a report in 2015 outlining the potential of SMRs. In addition to the benefits described above, the report identified some hurdles that still need to be overcome, such as testing the new technical approaches and overcoming institutional challenges, including regulatory barriers and the respective responsibilities of plant operators and host countries. ■
If Bill Gates succeeds in building the pilot plant and proving it can operate safely, that could spark a major shift in the industry.”
EPFL’s Soil Mechanics Laboratory has decades of experience working on the underground storage of nuclear waste. The Lab’s Director is Professor Lyesse Laloui: “Since 2006, we have been supporting the Swiss Nuclear Agency, NAGRA, to identify the most suitable site in Switzerland to store the country’s nuclear waste. We also have PhD students funded by the French Nuclear Agency and work with nuclear waste national agencies from Finland to Japan,” he said.
From a technical perspective, the storage of nuclear waste is an engineering process, however, because the decay of radioactive materials is so long — thousands if not tens of thousands of years — that it is imperative to find very stable and very safe solutions.
“In the history of humanity, the oldest constructions for which we have references are about 4,000 years old, so looking ahead 20,000 years is quite a challenge. It means that the physics that we put into our models to look at the behavior of the engineered barriers and storage clays needs to be very, very strong and our numerical and predictive tools need to be very capable. We’re trying to create predictive tools for non-tested conditions to understand how these materials will behave for tens of thousands of years,” Laloui continued.
In the history of humanity, the oldest constructions for which we have references are about 4,000 years old”
Tracking the gas leaks
One major current project is a Swiss National Science Foundation-funded study on compacted expansive clays that have a remarkable swelling capacity and swelling pressure, a self-healing capacity upon wetting, and very low permeability — making them suitable for engineering applications such as geosynthetic clay liners and nuclear waste geological storage. Laloui and his team are trying to understand the mechanical behavior of these geomaterials for their use as engineered geological barriers and to develop suitable corresponding modeling tools for more reliable design, which will lead to increased confidence in the environmental protection systems in place.
Another is a joint EU/Swiss project that falls within the scope of the European Joint Program on Radioactive Waste Management (EURAD). Gas generation and transport through a radioactive waste repository is an important issue for the geological storage and disposal of radioactive waste, so improving our understanding of gas transport through low permeability porous materials such as clay is considered a high-priority research area.
One day it is conceivable that the energy gap may be filled by non-radioactive nuclear fusion, but until that time, nuclear fission and its toxic by-product will remain with us on our journey to zero-carbon economies. While there is no such thing as zero risk, Laloui and his team remain confident that the long-term, safe and reliable storage of nuclear waste is possible. ■
Search Wikipedia for “nuclear reactor” and the first image that appears is of CROCUS — rather surprisingly, since this reactor is little known outside of scientific circles. It became a key element of nuclear-operations R&D when it was completed in 1983 and still plays an important role, especially in European research programs. “CROCUS is unique in Switzerland and extremely useful for both theoretical and applied research,” says Andreas Pautz, head of EPFL’s Laboratory for Reactor Physics and Systems Behaviour, where CROCUS is housed.
CROCUS is designed to be small and contained, with its capacity capped at 100 W, but it still has all the main features of a full-scale reactor. First, just like any other nuclear facility, it’s located in a highly secure area complete with restricted access, airlocks and radiation detectors. Second, to get to it you have go through a 14-ton pivot door and enter a room with concrete walls some 1.3 meters thick. The reactor core is covered by a 50-ton lid.
Redundant safety measures
The cylindrical core measures 60 centimeters across and 1 meter high. It contains an inner circle of 336 rods of uranium oxide enriched to 1.8%, along with an outer circle of 176 rods of uranium metal enriched to 0.95%. The rods are 1.2 meters high and cladded in pure aluminum. They are maintained vertically by two octagonal grid plates spaced 1 meter apart. The entire assembly sits in a bath of deionized water at 20°C, which serves as a moderator and reflector.
Like all nuclear reactors, CROCUS is designed to carry out nuclear fission in a controlled manner. The reactions are controlled by either adjusting the bath’s water level or deploying two absorber control rods. These rods contain boron carbide, a neutron-absorbing material, and are inserted into the top of the grid of uranium metal rods. In addition, the reactor has six independent shut-off systems that can plunge the reactor to subcritical conditions and halt the fission process in under one second. Each system alone can stop the chain reaction. Once that happens, it takes just a few hours for radiation levels to fall to the point where the core can be approached safely, with minimal precautions.
Noise in the core
All these features make CROCUS a valuable instrument for studying nuclear reactors. Pautz explains that it can be used to test new systems for measuring neutrons and gamma rays, check the results of computer-modeled reactor designs, collect data on nuclear reactions for building next-generation reactors, and investigate specific nuclear phenomena and their effects. For example, CROCUS was essential to the recently completed CORTEX project, which was carried out under the EU’s Horizon 2020 research program. “We began suspecting a few years ago that the fuel rods of a certain type of pressurized water reactor were vibrating, which could potentially disrupt a power plant’s operations,” says Pautz. “But since that was taking place in the core, we couldn’t observe what was going on. We therefore built a system for CROCUS that would make the rods vibrate in a controlled manner.” What the engineers did was make 18 fuel rods on the CROCUS core periphery oscillate in ranges of ±0.5 mm and ±2.0 mm, at frequencies between 0.1 Hz and 2 Hz. They also placed 11 neutron detectors both inside and outside the core to record signals emitted by the neutrons. “This experiment allowed us to characterize the fluctuations — or ’neutron noise’ — caused by the vibrating fuel rods, thanks to the measured signals. We used the findings to develop a monitoring system for reactor cores so that anomalies can be detected and pinpointed before they become a problem,” says Pautz. ■
CORTEX project, which was carried out under the EU’s Horizon 2020 research program
In the second half of the 21st century, the world will steadily transition away from fossil fuels as a new energy source comes to the fore: nuclear fusion. Despite their confusingly similar names, fusion and fission differ in many important ways. A fusion reaction doesn’t require rare metals. Nor does it generate long-lived hazardous waste. And it holds the promise of clean, safe and almost limitless energy.
All of this might sound too good to be true — which is perhaps why, in its early days, investors were reluctant to support this fledgling technology. But a number of major milestones have been achieved in recent years. The latest one happened on 21 December 2021, when engineers at the JET test reactor in Culham, UK, were able to sustain fusion reactions for five seconds and produce a record-breaking 59 megajoules of energy.
A star in a donut-shaped machine
To understand why this achievement matters, let’s go back to the basics. Nuclear fusion is nothing new. In fact, it’s the same reaction that’s powered the stars for over 13 billion years. But the idea of reproducing fusion on a much smaller scale — right here on Earth — didn’t get off the ground until the 1950s.
Of the various methods proposed at the time, the one that gained the most traction among the international scientific community was the “tokamak.” This name comes from the Russian acronym for a machine that confines plasma — the fourth state of matter, consisting of free-floating electrons — in a donut-shaped chamber using powerful magnetic fields. EPFL has its very own tokamak reactor, which also happens to be the largest research facility on the Lausanne campus . There are several other such reactors across Europe, including the one in Culham.
150 millions degrees Celsius
Nuclear fusion reactions happen routinely in stars due to their incredibly strong gravitational field. But on Earth, the plasma needs to be heated to 150 million degrees Celsius — 10 times hotter than the center of the Sun — to compensate for the much weaker force of gravity. And there’s another challenge: the plasma would severely damage the reactor wall if it touched it, so it has to be confined in the center of the reactor using powerful magnetic fields.
The plasma contains two isotopes of hydrogen known as deuterium and tritium. Atoms are fired at breakneck speed into the plasma where they collide with such energy that they fuse, producing helium atoms and releasing neutrons. This process also creates a huge amount of energy, as described by Einstein’s famous E=mc2 equation. The energy can be recovered and used to heat water, with the resulting steam fed into a turbine to generate power.
All sights set on 2035
The real challenge in this process is to keep the reaction going for long enough to produce usable heat. The five-second record set by the UK team marks a huge step forward, because it proves that sustained fusion reactions are possible. “This is great news for ITER, the giant reactor currently being built in the south of France,” explains Yves Martin, deputy director of EPFL’s Swiss Plasma Center.
ITER is a vast fusion reactor being developed at the Cadarache research center, not far from Marseille. Once complete, it will be the biggest tokamak on the planet. It’s slated to start producing plasma in 2025. ITER builds on the work done at JET and draws on the know-how of the SPC and a community of engineers and businesses from 35 different countries. The project aims to demonstrate the feasibility of generating clean energy from nuclear fusion reactions in a plasma, with the goal of sustaining a fusion reaction for around 500 seconds. Current predictions suggest they will achieve this in or around 2035. ■
EPFL and DeepMind use AI to control plasmas for nuclear fusion
Scientists at EPFL’s Swiss Plasma Center and DeepMind have jointly developed
a new method for controlling plasma configurations for use in nuclear fusion research.
EPFL’s Swiss Plasma Center (SPC) has decades of experience in plasma physics and plasma-handling methods. DeepMind is a scientific discovery company acquired by Google in 2014 that sits at the cutting edge of the research and development of advanced artificial intelligence. Together, they have developed a new magnetic control method for plasmas based on deep reinforcement learning, and applied it to a real-world plasma for the first time through the SPC’s tokamak research vessel. Their study has just been published in Nature.
Controlling a substance as hot as the Sun
To control plasma in the vessel, researchers at the SPC first test their control systems on a simulator before using them in the TCV tokamak. “Lengthy calculations are needed to determine the right value for each variable in the control system,” says Federico Felici, an SPC scientist and co-author of the study. “That’s where our joint research project with DeepMind comes in.”
DeepMind’s experts developed an AI algorithm that can create and maintain specific plasma configurations and trained it on the SPC’s simulator. This involved first having the algorithm try many different control strategies in simulation and gathering experience. Based on the collected experience, the algorithm generated a control strategy to produce the requested plasma configuration. After being trained, the AI-based system was able to create and maintain a wide range of plasma shapes and advanced configurations, including one where two separate plasmas are maintained simultaneously in the vessel. Finally, the research team tested their new system directly on the tokamak to see how it would perform under real-world conditions.
The SPC’s collaboration with DeepMind dates back to 2018, when Felici first met DeepMind scientists at a hackathon at the company’s London headquarters. There he explained his research group’s tokamak magnetic-control problem. “DeepMind was immediately interested in the prospect of testing their AI technology in a field such as nuclear fusion, and especially on a real-world system like a tokamak,” says Felici. Martin Riedmiller, control team lead at DeepMind and co-author of the study, adds that “our team’s mission is to research a new generation of AI systems — closed-loop controllers — that can learn in complex dynamic environments completely from scratch. Controlling a fusion plasma in the real world offers fantastic, albeit extremely challenging and complex, opportunities.”
A win-win collaboration
“We agreed to the idea right away, because we saw the huge potential for innovation,” says Ambrogio Fasoli, head of the SPC and a co-author of the study. “All the DeepMind scientists we worked with were highly enthusiastic and knew a lot about implementing AI in control systems.” For his part, Felici was impressed with the amazing things DeepMind can do in a short time when it focuses its efforts on a given project.
DeepMind also got a lot out of the joint research project, illustrating the benefits to both parties of taking a multidisciplinary approach. Brendan Tracey, a senior research engineer at DeepMind and co-author of the study, says: “The collaboration with the SPC pushes us to improve our reinforcement learning algorithms, and as a result can accelerate research on fusing plasmas.” This project should pave the way for EPFL to seek out other joint R&D opportunities with outside organizations. “We’re always open to innovative win-win collaborations where we can share ideas and explore new perspectives, thereby speeding the pace of technological development,” says Fasoli. ■
The collaboration with the SPC pushes us to improve our reinforcement learning algorithms.”
Can you tell us a little more about the Swiss Plasma Center?
The SPC is a unit within EPFL’s School of Basic Sciences. It’s Switzerland’s largest R&D center for plasma physics and fusion energy and one of the main fusion research centers in Europe. We have some 120 employees and 40 PhD students and postdocs who study all aspects of plasmas and nuclear fusion. In addition, through our ties with the EU, we’re a member of ITER, a global research project to demonstrate the feasibility of nuclear fusion from a scientific and technological standpoint.
What makes the SPC unique?
Our center plays an important role on several levels. We have a machine called a Variable Configuration Tokamak (TCV) that’s considered an essential tool for European research on plasma physics and configurations. The TCV is used both for the SPC’s domestic program and by EUROfusion — a European fusion-energy consortium that funds research on behalf of the European Commission’s Euratom program, and which receives substantial human and financial resources from the EU. The TCV is the biggest experimental facility at EPFL; scientists and engineers come to our school from across Europe and elsewhere in the world to run experiments on the TCV.
EPFL has another major fusion research facility at the Paul Scherrer Institute in Villigen. Here, scientists and engineers test conductors for magnets used by ITER and most of the world’s main fusion reactors. It’s the only facility of its kind globally and brings in a lot of R&D contracts.
In July 2021, EPFL opened an advanced computing hub for fusion applications. We’re now home to one of only five such hubs in Europe. The SPC runs the hub and leads its cross-disciplinary team consisting of researchers from EPFL’s Institute of Mathematics, SCITAS (which operates a high-performance scientific computing platform), the Swiss Data Science Center (a national center of excellence in big data) and the Laboratory for Experimental Museology (eM+).
Ambrogio Fasoli, Director of the Swiss Plasma Center © Gilles Gonin
How important is education to the field of nuclear fusion?
One thing’s for sure about our field: the research will span several generations. We’re not far from our goal, but it’ll still be a while before we’re fully capable of controlling fusion reactions and then generating and distributing fusion energy. The next generation of scientists and engineers will have to take up the challenge, and education will be key to getting them ready.
At the SPC, we play a central role in education by training cohorts of young physicists and engineers from Switzerland and elsewhere in Europe, so as to prepare them for ITER and whatever projects may come next. The numerous classes given by our center are highly appreciated, and we’ve developed a series of MOOCs — the world’s first on plasma and fusion — which have been taken by nearly 20,000 people. That’s a record for the topics of plasma physics and nuclear fusion, and we’re proud.
How can nuclear fusion help provide responses to today’s environmental and energy challenges?
Nuclear fusion won’t be the only solution to the clean-energy problem, but it can fundamentally change how baseload power is produced. It emits no greenhouse gases, is not dependent on weather conditions, and is completely different from nuclear fission — meaning it’s safe and doesn’t produce long-lived radioactive waste.
Of course, we fully support renewable energy like wind and solar power, but their potential is inherently limited, due to their intermittent supply and storage requirements, for example. This makes them ill-suited to providing a steady supply of baseload power. Fusion fuels on the other hand can be distributed widely and in a nearly infinite supply, and they can last thousands of years, thus enabling continuous energy production. Nuclear fusion can provide clean, safe and stable baseload power for the future of humanity.
Nuclear fusion won’t be the only solution to the clean-energy problem, but it can fundamentally change how baseload power is produced.”
When will nuclear fusion be able to replace nuclear fission or even fossil fuels like coal, oil and natural gas?
We don’t have an answer to that yet. The ITER test reactor is under construction and should be ready for experiments by 2027. The SPC is also a member of the DEMO project, which aims to demonstrate that fusion energy can be distributed viably on the market. We will incorporate the lessons learned from ITER into the design, engineering and operation of the DEMO reactor.
We should be able to prove that fusion energy can be produced and sold starting in 2050. It’s an ambitious goal, but a realistic one, and it doesn’t preclude innovations or breakthroughs in the meantime that could speed up the process.
How has your center been affected by the shelving of the framework agreement between Switzerland and the EU in spring 2021?
The end of the framework agreement made it unclear whether Switzerland can continue to participate in Horizon Europe, which is the EU’s ninth funding program for research and innovation, or in Euratom, which is the European organization for developing industrial applications for nuclear fission and fusion — and our gateway to ITER. This could have major consequences for our scientists and engineers working on ITER, but also for the Swiss manufacturers who are developing and supplying parts for the ITER thermonuclear reactor.
There will obviously be a large financial impact, but the ramifications will go much further. Switzerland’s exclusion from these programs is a problem because we make an important contribution to R&D on the European level. Being part of this community lets us work with world-class scientists and engineers from other countries, and lets those in other countries work with ours. We can do much more as a team, pooling the knowledge of top experts through an integrated program, than we ever could individually.
What options are available for getting around this obstacle?
With regard to EUROfusion, which is Euratom’s right arm, Switzerland and the UK have become associate partners through Germany. Technically, we’ll only be observers with no voting rights on EUROfusion’s governing bodies, but we’ll be able to fund and take part in R&D projects. The other question mark is Switzerland’s participation in the ITER project itself. Bernard Bigot, the ITER Organization’s Director-General, underscored Switzerland’s importance to the project last December when he came to EPFL to speak with Martina Hirayama, who is Switzerland’s State Secretary for Education, Research and Innovation, EPFL President Martin Vetterli and me. We’re now thinking about setting up a cooperation agreement between the ITER Organization and EPFL as a research institute. Such an agreement would have to be approved unanimously in writing by all ITER members, including the EU. ■
The giant fusion reactor
“The ITER project aims to bring together the best scientific and engineering skills from around the world in order to demonstrate the feasibility of using nearly self-sustaining hydrogen plasma to generate 500 MW of thermal energy. The SPC is widely recognized as a global center of excellence in fusion research and has resources that can’t be found anywhere else. It’s therefore essential for ITER to be able to keep working with EPFL and the SPC in the coming years — irrespective of Swiss-EU politics — and continue the partnership that has been so beneficial to both sides for over 12 years.”
Bernard Bigot, general director of ITER
Earlier in this special feature, Marc Vielle described the future of nuclear as being “far from a bed of roses.” Andreas Pautz took a different view, arguing that he couldn’t see “how we can do without it.” François Maréchal, who heads the Industrial Process and Energy Systems Engineering (IPESE) group at EPFL’s Valais Wallis campus, is categorical in his belief that reviving the nuclear industry is a “very bad idea” and that it has no part to play in the transition to clean energy. He points to the unresolved headache of managing radioactive waste as one reason for this view. But he also raises another factor: time. “It’s just too late for nuclear power to bring us anywhere close to reaching our climate goals by 2050,” he explains.
There’s no shortage of evidence to support Maréchal’s claim. Plans to build new EPR nuclear plants in Europe have been beset by problems, with Flamanville (France), Olkiluoto (Finland) and Hinkley Point (UK) all facing delays and cost overruns. In China, meanwhile, Taishan — the world’s first “new generation” EPR plant to come online — was shut down last summer after less than three years of service because of technical problems. “Add in political decisions — the closure of Germany’s remaining plants this year and the moratorium on nuclear power in Switzerland — and it becomes clear that, in Europe at least, getting a new plant up and running before 2050 is out of the question,” says Maréchal.
A palette of options
Maréchal cautions against pinning all our hopes on nuclear power. “We need to invest as of today in an array of decentralized solutions based on renewables rather than putting all our eggs in one basket,” he explains. “The share of nuclear power in the future energy mix is a side issue. What really matters is managing energy supply and developing the right power distribution and storage infrastructure. And we’ll need people with the skills and know-how to build and operate these systems.”
Looking ahead, a growing percentage of our electricity will almost certainly come from intermittent sources. This will render the idea of cross-border energy trade obsolete: countries will have no interest in buying in power from their neighbors at the height of summer when their own plants are running at full capacity. Moreover, commodity price volatility, exemplified by the gas price surge in 2021, will signal the end of the era of cheap energy.
“This situation calls for a holistic approach,” explains Maréchal. “We have to cast a wide net and consider every part of an energy system: production, storage, distribution and consumption. There’s a lot of research happening in all these areas, but it needs to be merged together. And this doesn’t apply just to electricity: it concerns all forms of energy, including heat — and how we will spare it.”
The share of nuclear power in the future energy mix is a side issue”
Hydrogen is a rising star
Exploring new avenues, expanding our R&D efforts, and building innovative systems will all be crucial. And EPFL intends to play its part, with dozens of research labs already engaged in energy-related programs, both directly and indirectly, studying topics ranging from silicium and perovskite solar cells biomass recovery to concentrator photovoltaics, fuel cells and electrolysis, hydropower, geothermal energy, CO2 storage, nuclear engineering, plasma physics and more.
Some EPFL labs are working specifically on hydrogen technology and storage, tackling the subject from different angles. The ability to produce clean hydrogen from renewable electricity or biomass, and even transform it into other, more easily storable chemical compounds, will play a key role in our energy future. The gas can be used to make fuels such as synthetic kerosene, thus reducing our reliance on oil. It can also provide a source of clean electricity, generated on-demand inside fuel cells for use in stationary systems as well as in trucks and other mobile applications. The technology is advancing at a rapid pace, with some systems already reaching the market. There’s no doubt that hydrogen production, storage and distribution will be key components of tomorrow’s energy mix. Whether new nuclear plants form part of that mix remains to be seen. ■
Swiss-Energyscope: Supporting informed decision-making
The sheer complexity of energy systems can leave policymakers feeling lost — and potentially vulnerable to influence
from powerful lobby groups. With so many factors to consider, especially the timescales involved in building power grids, these kinds of decisions can push the limits of human reasoning.
Swiss-Energyscope is a simulator developed at EPFL to address precisely this problem, giving policymakers insight into issues around the energy transition and letting them compare options and their long-term consequences. The software incorporates a vast array of variables, which can be adjusted to generate 15- and 30-year models. It is available online (calculator.energyscope.ch) and backed, among others, by the Swiss Federal Office of Energy and Swiss Competence Centers on Energy Research; the program is updated regularly to reflect technological developments and the latest real-world energy consumption data. In the future, it will be extended to countries other than Switzerland.
One of Swiss-Energyscope’s key benefits is that it allows the consequences of decisions to be assessed without a significant investment in cost or time. For instance, if you slide the bar for nuclear power to a higher level, the software calculates how building new nuclear plants would affect Switzerland’s energy supply and carbon emissions. ■