To the editors:
In his essay, Daniel Jassby covers recent advances and results in two well-known fusion machines, the National Ignition Facility (NIF) in Livermore, California, and the Joint European Torus (JET) near Oxford, England. Both machines have performed recent runs that have set new records, the NIF in inertial confinement fusion (ICF) and the JET in magnetic confinement fusion (MCF), although the magnitude of the advances has been much greater in NIF. Jassby draws the conclusion that, “[b]ased on actual performance, ICF appears to be a far more likely candidate than MCF as the basis for a power plant.”
I do not believe that his essay justifies this claim. The new results are certainly interesting, but they have little bearing on a future power plant. Consider: What is the purpose of a hydroelectric power plant? Is it to produce water? And then, what is the purpose of a fusion power plant? Is it to produce fusion? If the purpose of a power plant is to produce electricity, then this essay does not support its conclusion. In fact, the essay is concerned mostly with plasma physics. Jassby cites outcomes from the two fusion machines:
- NIF: 1.9 megajoules (MJ) of laser energy in, 1.3MJ fusion energy out;
- JET: 33 megawatts (MW) of heater energy in, 11MW of fusion out.
Note the different units reported by the two teams: these have no effect on the resulting dimensionless ratio. In terms of Q, the fabled break-even measurement, these results are
- NIF: 1.3 / 1.9 = 0.68,
- JET: 11 / 33 = 0.33.
It would seem NIF is twice as good as JET.
But these values do not account for the total energy consumption of the machine. NIF is powered by an enormous laser, while JET uses large copper magnets. Both are power-hungry. The fusion energy output is mostly in the form of neutrons, and converting them to electricity will be about 35% efficient, at best.1 When one considers not the physics, but the electricity, a very different picture emerges:
- NIF: 400MJ in, 0.45 out,
- JET: 700MW in, 3.9 out.
The equivalent Q is sometimes known as “Q engineering” or “Qeng.” It must be greater than 1 for the device to be a power plant. In this case,
- NIF: 0.45 / 400 = 0.001,
- JET: 3.9 / 700 = 0.006.
By this measure JET is six times as good as NIF.
Jassby details potential improvements to both designs. For NIF, it might be possible to replace the ~1% efficient lasers with other drivers. For heavy-ion drivers (HIF), I have often seen the figure of 30% mentioned. Likewise, JET could replace its magnets with superconducting versions, which would reduce its power input to about 100MW. If we consider both of these,
- HIF-NIF: 0.45 / 13.5 = 0.03,
- Super-JET: 3.9 / 100 = 0.04.
It needs to be pointed out that while superconducting MCF experiments have existed since 1965, no alternative driver for ICF is in use and I cannot find serious plans to build one. If I am interpreting NIF results correctly, it appears these alternative drivers might not be able to deliver the extremely precise pulse timings used in experiments. If lasers are the only solution, new technology may provide as much as 10% efficiency, giving a Qeng figure a little under 0.01. To make either work will require significantly better physics, but the path forward is well understood for MCF, while ICF’s performance remains murky.
For the true energy wonks, one might suggest that the purpose of a power plant is not to produce power, but money. In that case there is a final consideration, the fuel burnup. If a power plant throws away most of its fuel, it will never be economically viable. For a variety of reasons, ICF devices only burn the majority of their fuel if they operate at much higher gains. In a famous 1972 paper in Nature, John Nuckolls et al. spell out the reasons in detail and suggest a minimum Q around 50 is required.2 Modern results suggest this is closer to 100.3 In contrast, MCF devices burn up most of their fuel at much lower Q values: > 5 would be ideal. If we take those amounts as the yardstick for commercial success, Super-JET’s 0.04 is much closer to 5 than HIF-NIF’s 0.03 is to 100.
For all these reasons, I would suggest that MCF is closer to commercialization than ICF. I believe the relative lack of development in ICF compared to MCF reflects this. Many dozens of MCF projects are underway, including the International Thermonuclear Experimental Reactor, the world’s largest experiment. The number of ICF devices, in contrast, can be counted on one hand, and most are primarily military devices for testing bomb physics.
Still, saying MCF is closer to a power plant than is ICF is like saying a 20-foot ladder is closer to landing you on the moon than is a 10-foot ladder. What these numbers really illustrate is not which approach is better, but the remaining gulf that has to be crossed—after over 70 years of effort in this field, the two best devices in the world are still orders of magnitude away from being useful.
Maury Markowitz
Daniel Jassby replies:
Maury Markowitz’s letter concerns power balance issues for reactors, which are at best tangential to my essay. As related in the introductory paragraphs, my essay is concerned with the scientific feasibility of thermonuclear plasmas, not with input and output energies of hypothetical reactors. Nevertheless, I must point out that Markowitz fails to differentiate beam-thermal from thermonuclear reactions, a vital distinction discussed at some length in my essay. Only the latter reactions can be scaled to the high Q-values needed for electrical energy production.
Markowitz’s analysis is based entirely on two recently publicized shots of the NIF and JET. In the JET shot that he analyzes, at most one-fourth of the fusion power was thermonuclear. If Markowitz wants to use data that could be extrapolated to practical reactor conditions, then his various performance parameters for JET should be divided by a factor of 4. Hence JET’s thermonuclear-based fusion output would be 3MW instead of 11MW and its electrical power output 1.0MW instead of 3.9MW. Then the thermonuclear-based Qeng is 0.0015 for JET and 0.01 for Markowitz’s Super-JET, compared with 0.001 and 0.03 for the respective ICF cases. The comparison is mixed, and all values are tiny, so that nothing can be concluded from this evaluation.
Markowitz claims that MCF is “closer to commercialization” than ICF, citing the far higher numbers of MCF projects compared to just a few ICF facilities. There are indeed dozens of experimental tokamaks worldwide because resistive-coil versions are relatively easy to construct and give impressive plasma performance at acceptable cost. But is that a meaningful criterion? The number of ICF and MIF (magneto-inertial) projects that have actually used the critical tritium fuel is many times higher than the number of MCF projects that have used tritium—exactly two in 70 years!
Over the decades there have also been hundreds of plasma pinches and focuses, dozens of magnetic mirrors, and countless electrostatic fusors, all of which have been abandoned as viable approaches to fusion reactors,4 although some are still operating.5 It is possible that tokamaks, too, may one day join the collection of fusion zombies.
Markowitz’s final paragraph appears to be in agreement with my essay’s conclusion that practical fusion power reactors of any type are a distant prospect.
