Fusion power has been 30 years away for about 50 years now, and at this point the joke tells itself. What doesn’t get said as often is where the hold-up actually sits. Getting hydrogen hot enough to fuse is a solved trick, and there are machines doing it today. Keeping the plasma from chewing up the inside of the machine it lives in, year after year, with nobody standing there nudging it, is a job nobody has finished.<\/p>\n
Britain’s flagship machine spent its latest run on exactly that. The UK Atomic Energy Authority<\/a> closed out the fifth experimental campaign on MAST Upgrade, at its Culham site in Oxfordshire, and published the results on August 6. The run turned out more than 1,100 plasmas, the highest pressure the machine’s ever held without the plasma going unstable, and a control technique nobody had pulled off before, which comes down to watching the thing glow.<\/p>\n UKAEA hasn’t published a number for that pressure, so I can’t tell you how it stacks up against anything else, and I’d rather say that than dress up a record with no number on it. Nobody’s claiming this machine makes more energy than it eats, either. It doesn’t, and it was never built to.<\/p>\n A tokamak holds its plasma inside a magnetic cage, and the plasma pushes back. Pressure builds at the outer edge until the edge lets go all at once, and that’s an Edge Localized Mode. UKAEA describes them as sudden bursts that can throw out as much as a tenth of the plasma’s stored energy in one go. Do that a few thousand times and you’ve chewed up the inner wall and the exhaust hardware.<\/p>\n You should care about somebody else’s wall for the same reason you care about a timing belt. Parts that need replacing mean downtime, and downtime lands in what the electricity costs. UKAEA’s direct about it: left alone, ELMs would push maintenance high enough to keep fusion from competing on price. The fission side already shows what that looks like, since a generator rotor that failed twice<\/a> has held a Finnish reactor at 735 megawatts instead of 890 ever since.<\/p>\n The team went at the burst problem four separate ways in the same run. Two of them, Quasi-Continuous Exhaust mode and Resonant Magnetic Perturbations, build on earlier work with 3D magnetic coils<\/a> on the same machine. RMP uses those coils to hold the edge pressure down far enough that it never builds to a burst. QCE is pretty much what it sounds like, an edge that leaks a little all the time instead of saving it up.<\/p>\n The other two are Quiescent H-mode and I-mode, and they’re confinement regimes rather than suppression tricks. UKAEA’s description is that they hold energy in better while keeping the edge under control, without setting off the big bursts.<\/p>\n Getting all four inside a spherical tokamak counts for something. A spherical tokamak is the conventional design squashed inward, closer to a cored apple than a donut, and plasma behavior in there doesn’t automatically match what other machines see. James Harrison, head of MAST Upgrade science at UKAEA, said the results “genuinely shape the design of future fusion power plants.” You’ll hear a version of that from every lab. This one has four regimes and a new diagnostic sitting behind it, so I’d let it stand.<\/p>\n You can’t stick a ruler in it.<\/p>\n UKAEA puts the plasma in this machine at about 54 million degrees Fahrenheit (30 million Celsius), and anything you push into that either evaporates or poisons the plasma with whatever it’s made of. So the machine gets read from a distance. Position matters because the plasma’s got to sit where the magnets expect it, and a small drift puts the heat somewhere nobody designed for.<\/p>\n MAST Upgrade runs a shape with an exhaust region at the top and another at the bottom. Those are the divertors, and a divertor is a section of wall the machine deliberately aims the escaping heat and particles at, so the rest of the wall doesn’t take them.<\/p>\n Deuterium is hydrogen with an extra neutron in it, and it’s half the fuel in there. It glows in visible light where it reaches those exhaust regions, and the team measured that glow coming off the upper and lower outer divertors. Basically, when the two don’t match, the plasma’s moved. UKAEA says the technique catches minute positional imbalances in real time, while the shot is still running, and calls it a world first.<\/p>\n A power plant can’t have somebody watching a screen and nudging the magnets by hand, so it’s got to correct itself, and that puts everything on the instruments being right.<\/p>\n Instruments are their own genre of problem. Last week a single failed nuclear instrument at a plant in Illinois<\/a> convinced half the reactor protection system that the control rods were moving when they weren’t, and the crew shut that unit down by hand at 4:31 in the morning. A position signal that comes straight off the light the fuel gives away, with nothing mechanical in the chain, is a pretty attractive thing to have.<\/p>\n The campaign’s other result is about where the heat ends up. Put a little nitrogen into the edge of the plasma, the team found, and the plasma throws most of the exhaust power back out as light. Light leaves in every direction at once, so instead of a concentrated load landing on one strip of metal, the energy gets shed through the volume before it reaches the wall or the divertor. Peak heat flux drops. So does the wear on the surfaces underneath.<\/p>\nSo what’s an ELM?<\/h2>\n
Four stable regimes came out of one campaign<\/h2>\n
So how do you find a plasma you can’t touch?<\/h2>\n
Nitrogen turns exhaust heat into light<\/h2>\n