The Star Builders
Page 20
It’s an exciting period of nuclear innovation by governments that have deep pockets and rich nuclear expertise. For its part, Eurofusion, the club of European countries pursuing fusion, is also considering stellarators as a path to net energy gain—in particular, due to the astounding success of Wendelstein 7-X.20
And the private sector fusion firms claim that they’re hard at the heels of the big laboratories, even if the data suggest they’re behind right now. Most are promising to get to net energy gain (or, at least, net energy gain conditions) in the 2020s or early 2030s—well before ITER. General Fusion and Tokamak Energy see it happening by 2022, Lockheed Martin by the 2020s (revised from 2017), First Light Fusion by within the 2020s, and TAE, to stay on track, will need to have reached net energy gain and commercialized its technology by 2024. If we take these claims seriously, the star builders’ race for net energy gain is close to finishing—but wide open.
Looking at the sweep of history, the progress made to date, and the promises being made from all quarters, it seems certain that net energy gain from fusion can and will be achieved. It could be this year. It could be next. Maybe it’s thirty years away, like the joke about fusion says. For what it’s worth, I think it will be sooner than that.
John Lawson showed us that net energy gain was theoretically possible. Extrapolation from magnetic star machines and interpolation from secret nuclear tests have shown us that it’s experimentally possible. Just how fast net energy gain arrives will be determined by how badly nations and entrepreneurs want to see it happen.
But net energy gain is coming.
After Net Energy Gain
The latest star machines are tantalizingly close to demonstrating net energy gain from nuclear fusion reactions. So let’s try a thought experiment: Imagine that net energy gain is achieved. Not just a small gain of one, but a big gain—thirty or more. A gain that is on the verge of what is needed to build a power plant. It could have happened with NIF, on one of the machines being built by a start-up, or maybe with ITER, when it’s finished. For this thought experiment, it doesn’t matter who. Now ask yourself a question: What next?
It’s a question the star builders are increasingly asking themselves too: How will they take nuclear fusion from physics experiment to power plant?
As we sit in his office at Culham, Ian Chapman gives me a rundown of what needs to happen for tokamaks to successfully deliver energy to the grid. “There are five big challenges,” he begins. The first is plasma that is ten times hotter than the center of the Sun: “JET is operating at 150 million degrees, so we know how to do that.” This goal has been met, and any machine that gets net energy gain will have achieved it too. The temperatures used in inertial confinement fusion are typically lower than that, and NIF is just shy of what it needs for net energy gain. Ian lowers the thumb of his outstretched hand, four points to go.
“Second: How do we get the heat out of that big ball of gas with turbulence in the middle?” Once net energy gain is demonstrated, the energy produced must be extracted and used to turn water into steam. The steam will drive a turbine to produce electricity. Star builders have ideas about how to do this, but it’s hard to make progress without a net-energy-gain reactor to try them on. For tokamaks, the first stage, getting heat energy out, is the hardest. “We sweep it down to a sacrificial surface at the bottom,” Chapman continues. This is the divertor, a part of the tokamak designed to take a beating that’s more intense than what the space shuttle receives during reentry. The bigger the tokamak, the more intense the heat. ITER is designed right at the physical limit of what solid materials in conventional tokamaks can endure. For inertial confinement fusion, the geometry of the reactor is much more simple and getting the heat out is too, so it’s less of a concern. Ian lowers a finger, ticking exhaust heat off his list.
“The third challenge is neutrons: we [will] have the most intense source of neutrons on Earth.” No one knows how surfaces will react to the number and energy of neutrons that a high-gain star machine will release—because building one is the only way to produce neutrons in sufficient numbers and with sufficient energy. Excessive neutron bombardment can make the hardiest materials crumble. Then there’s the potential for excessive radioactivity if neutrons are absorbed by the wrong materials. Ian tells me that choosing the right materials for the reactor will make the difference between a plant that lasts four years and one that lasts forty. Due to the irradiation of the chamber by neutrons, repairs will be very hard to make, which could lead to reactors shutting down for long periods of time.
“The fourth challenge is breeding tritium.” Tritium doesn’t last long in nature because of its short half-life of twelve years. It has to be created from lithium. When lithium captures a neutron, it turns into tritium. Lithium is useful for many other items, like batteries, but Ian stresses that not much lithium is needed for tritium breeding because fusion has such a wildly high energy density. “For the rest of my lifetime,” he says, meaning the energy he will use throughout his life, “I need a bathtub of water and the lithium you’d find in two laptop batteries.” So the problem is not obtaining lithium, it’s that no one’s ever demonstrated breeding tritium from fusion neutrons and lithium on the scale needed—there’s never been a reactor producing enough neutrons to try. After a few years of operation, the plan is that ITER will test out a lithium blanket around the reactor chamber, a first step in practical tritium breeding, but it won’t be a production-grade system. Almost all digits down, Ian Chapman has just one finger in the air. We’ll return to his fifth and final challenge for fusion later.21
Star builders today are thinking about how to conquer these challenges in a way that their predecessors just didn’t need to. With net energy gain promised within fifteen years (not thirty!) by almost everyone, the need for solutions to the engineering challenges that will come with a fusion power plant is focusing minds.
Ian Chapman is taking the challenges very seriously indeed. Culham’s current spherical tokamak, MAST Upgrade, is being used to explore the physics of the hot plasma exhaust that, in advanced tokamaks, will safely carry energy away from the hottest part of the plasma. They have a new Materials Research Facility tasked with looking at how materials will behave under the extreme conditions found in a nuclear fusion reactor, especially the constant bombardment by neutrons (there’s a similar international facility funded by Japan and the EU). Also at Culham, a Fusion Technology Facility will try out different materials under all of the other stresses of fusion (electromagnetic, mechanical, and thermal) to see what works best. And there are tens of millions of pounds put aside for a Hydrogen-3 Advanced Technology Centre dedicated to the challenges of handling and breeding tritium. Similar facilities are being established by star builders across the world. Robotic arms are already being used for remote handling of irradiated materials on JET, helped by a new dedicated robotics facility that is trying to extend and improve the technology. Robots that can go in and fix problems with the reactor quickly can help prevent long periods of shutdown.22
It’s one thing to overcome these challenges in isolation in specialist facilities, but it’s quite another to achieve them together in a working, high-gain fusion reactor.
The magnetic fusion star builders are already planning to build another tokamak after ITER that will overcome the final barriers to commercialization and act as a demonstration power plant. This will deliver energy to the grid, and all that’s necessary to achieve it: heat extraction (ultimately to drive a turbine), tritium breeding, robotic maintenance, and neutron-compatible materials. The power-to-the-grid reactor will be called DEMO, for demonstration power plant.
DEMO is more of an idea than a certainty, and the design is a work in progress. Star builders are waiting to see how ITER’s first years of operation go before deciding what to keep and tweak in DEMO. What most designs suggest is that DEMO, if it gets the go-ahead, will generate five hundred megawatts, equivalent to a modestly sized nuclear fission power plant, and will have a
gain of at least thirty times the energy put in.
DEMO faces many challenges. It probably needs to be slightly bigger than ITER; it needs to run with higher-density plasma; it needs to run for hours at a time, meaning that disruptions must be kept under control; and it needs to be completely self-sufficient in tritium.23
The laser star builders have thought about a demonstration power plant too. Inertial fusion faces more or less the same challenges as magnetic fusion, but one major difference is that inertial fusion reactors must rapidly repeat. Dr. Nick Hawker described this as First Light Fusion’s biggest challenge—they’ll need to have a shot as frequently as every five seconds.
NIF’s laser currently manages four hundred shots per year. To be economical, laser fusion for energy will need to run with ten shots per second. The reason many repeats are required is that each capsule explosion on NIF doesn’t release that much energy. For power generation, there needs to be a lot more energy coming out per second, and the energy needs to be released at a constant rate. Only by rapidly firing the laser at targets to produce mini-explosions, a bit like a gas engine in a car, can laser fusion devices hope to be commercially viable.24
With the current technology on NIF, this is impossible to do without breaking the laser. The optics need hours to cool down after each shot. Not to mention that the flashlamps take time to charge up and are horribly inefficient: of the 400 megajoules drawn from the grid, just half a percent (1.8 megajoules) currently make it into the laser beam.
Since NIF was designed several decades ago, laser technology has moved on, and the flashlamps that energize NIF’s 192 laser beams have been surpassed by diode-energized lasers. The newer diode lasers are around twenty to forty times more efficient and produce less waste heat. But while they can fire an impressive ten times a second, they’re a long way from being able to squeeze enough energy into each shot.25
Then there are the fiendishly complex targets that inertial fusion typically requires. For NIF, the cost of these needs to tumble from hundreds of thousands of dollars to just a couple of cents to make commercialization viable—remember that ten will be used up every second. When I raised with scientists at NIF how improbable that seems, they pointed out that modern manufacturing is effective at making precision-engineered products that cost cents—for example, bullets.
Livermore’s plans for a power plant based on laser fusion found expression in a prototype design called LIFE (Laser Inertial Fusion Energy). The program to design LIFE was canceled when scientists couldn’t get NIF to produce as much energy gain as they had predicted theoretically in the early 2010s. If NIF hits net energy gain, especially if it does so before any other star machine, then you can expect that the plans for an inertial fusion power plant will be quickly dusted off and scientists will focus on solving some of the fearsome challenges. But, for now at least, LIFE is dead.26
Notably, none of the challenges that face either the magnetic or laser approaches to fusion are true showstoppers: they’re likely to be solvable with sufficient investment of resources and engineering skill. After all, until the current generation of star machines, no device had ever been the hottest place in the solar system or reached the density of the Sun’s core.
The hard challenges of commercializing fusion and using it to save the planet don’t end with engineering the reactor itself, however.
A major concern of star builders is that DEMO and LIFE, if they do happen, will come far too late for climate change. DEMO is scheduled for twenty years after ITER comes online—that is, in the 2050s—when many nations have said they’ll already have achieved net zero carbon emissions. A LIFE-like reactor will probably only get into the planning stage after, and only if, NIF hits net energy gain. And neither LIFE nor DEMO will be selling electricity, even if they do supply it to the grid. They’re demonstration power plants, not the first commercial fusion power stations.
The new fusion entrepreneurs are absolutely, painfully right that the pace of progress in fusion energy is going to be too slow to help save the planet unless star building gears up. Ian Chapman told me that fusion was better late than never. While that’s true, if the star builders really want to ensure fusion energy is ready in time to be useful in preventing the worst effects of climate change, they need to step it up a notch.
Ian Chapman said that with the current funding and risk profile of the international collaborations on magnetic fusion energy, power wouldn’t be getting to the grid until 2050. From that, he sees rapid deployment, “the same growth rate as fission had, about 30 percent year-on-year, and the same scale of plant and cost of capital.” Even with this apparently rapid rate of deployment, starting from 2050 means that fusion would only power half the world’s electricity needs by 2083. With few other options, policymakers may be forced to continue with a mix of energy production that includes fossil fuels, as well as nuclear fission and renewables.
The start-ups want to get fusion power onto the grid significantly more quickly. Tokamak Energy told me that star power must be deployed at “huge scales” during the 2030s. Nick Hawker has worked backward to get First Light Fusion’s plan of action: “If we want to contribute to net zero by 2050 we need to be building plants, multiple, in the 2040s. And the first of a kind has to be built in the 2030s. Which means the physics problem has to be solved in the 2020s.” If they’re right, fusion plants will have to appear very quickly once net energy gain is achieved.
Can We Afford to Do Fusion?
The progress toward net energy gain, and more recently, the challenges that stand in the way of fusion power stations, all cost money. All of which means it’s high time to find out what UK Atomic Energy Authority CEO Ian Chapman’s fifth challenge is. “This one matters to me a lot,” he says, more seriously. “If you do all of them, but the cost of electricity is chronic [prohibitive], then we have fusion, but it doesn’t penetrate the market.” All of the work, the battles against plasmas, the decades of innovation—they won’t matter if the economics aren’t right.
Money—to fund people, machines, or other resources—can speed up progress toward fusion. Conversely, lack of money can slow it down. When people ask how long it will be until fusion is achieved, they may as well be asking how much more money needs to be spent to get to a working reactor. The answer to how close fusion is as a power source depends on our collective appetite to see it happen.
Fusion is an expensive endeavor, partly because star machines seem to demand ever bigger scales—and bigger scales incur disproportionately bigger costs. ITER is going to be bigger than JET, DEMO is likely to be bigger yet. Initial estimates of the total cost of ITER were around €5 billion ($6.7 billion), but that had shot up to €20 billion (about $26.6 billion) by 2016. Even ITER insiders agree that scale is a problem. Sir Steve Cowley, the former CEO of the UK Atomic Energy Authority, has said, “I fully support ITER because we have to do a burning [self-sustaining fusion] plasma. But commercial reactors will need to be smaller and cheaper.”27
NIF was supposed to cost $1.4 billion, but cost overruns took the total price tag to $4.1 billion.28 Rough plans for a successor to NIF wouldn’t need a bigger building, because of advances in laser technology, but would probably need to scale up the laser energy, and this adds to the cost. “That’s why the NIF laser is not viable,” First Light Fusion’s Nick Hawker told me, when we were discussing commercialization. “The NIF laser was over $1000 per joule of energy, and our gain machine is going to be between four and five dollars per joule of energy. Even at that price, it’s a problem.”
Of course, first-of-a-kind reactors have research and development costs that don’t apply to mature technologies. But star builders still think that capital costs will dominate commercial fusion reactors. “The actual fuel costs will be small,” Professor Sibylle Günter told me. She listed the two biggest costs as the initial investment for the device and buildings, and the maintenance of the plant. This could be a problem, as finding funding for capital intensive power plants, with payoffs
over decades, isn’t easy. The UK government, for example, has struggled to find private financing for a mammoth three-thousand-megawatt fission plant at Hinkley Point in Somerset.
Here’s where the new star builders have a killer argument: if their machines work, they’ll be smaller and less capital intensive. “A small, modular device means more can be built off-site in a factory, and that’s a big factor to driving cost down,” Tokamak Energy’s CEO, Jonathan Carling, told me. Tokamak Energy has published academic research suggesting that the need for scale in tokamaks may be a false supposition. They hope that their spherical tokamaks can sidestep the need for ever bigger machines. The huge price drops that can be achieved with small modular devices is a well-known phenomenon—Moore’s Law, for microchips, is one famous example. Solar cells are another: they became more than three thousand times cheaper between 1956 and 2019. First Light Fusion is keeping costs down by relying on off-the-shelf technologies whenever possible.29
However, it’s a big “if.” What if we have to rely on government attempts to do fusion, which so far have utilized huge machines? There’s no doubt that the price tags for JET ($2.6 billion), NIF ($4.1 billion), and ITER ($22 billion) are high, but maybe that’s not intolerable. Those sums aren’t so different from the cost of the Large Hadron Collider at CERN ($5.3 billion), the Square Kilometre Array ($1 billion), and the most expensive big science project ever, the International Space Station ($120 billion). The next generation of particle collider will probably cost $25 billion. Even with their high price tags, these are all extremely worthwhile programs that will enhance our species’ existence. But while all will deliver new scientific knowledge, star builders would say that only the fusion machines are capable of delivering a new energy source. And a new energy source is desperately needed.30