Fusion is much with us these days; late and soon. As the fact that intermittents cannot solve the Gordian Knot starts to sink in, billionaires, journalists, and other experts have turned in desperation to the old Holy Grail. The good news is fusion is getting private funding, so at least it will have a shot if it really is economic. Under ITER style funding, even the best technology would die.1
Fusion faces a number of extremely difficult, unsolved, technical problems: damage to the first wall, enormous heat transfer rates through that wall, handling large amounts of pure lithium, activation of the whole mess making maintenance and repair as difficult as fission. But the problem I have with fusion is economic. Suppose you solve all these daunting technical problems. You are still faced with immense parasitic loads: plasma heating via microwaves, magnet power, vacuum pumps, cryogenic systems. This implies that the fusion gain, the ratio of the heat in to the reactor to the heat out that is needed to generate net electric power can be a factor of 40 or more, Figure 1.
Figure 1. Princeton calc of required fusion gain. Scale on left is ratio of power out to power in. Less than 1.00 is an electricity sink.
Figure 1 is hard to read. On the left is ratio of electrical power out to electrical power in. We need to come out ahead; say we want 1.05. Pick a thermal efficiency. 0.45 (the red line) would be a reasonably achievable number. Go across at 1.05 until you hit the red line, and then go down to the dashed line. You will be about at 34. That's the fusion gain you need.2 The paper is from the Princeton Plasma Physics Lab.\cite{menard-2016} These guys know their stuff, but they are also selling a prototype. Most of the assumptions underlying Figure 1 are optimistic.
Fusion is not self-sustaining in the sense that fission is. Each fission produces more neutrons than you started with which produces more fission, which results in a chain reaction. The problem is keeping this process under control; but thanks to the existence of delayed neutrons that is not difficult. Fusion has no counterpart. It just creates rapidly cooling heat.
Standard fusion is based on burning deuterium and tritium, the D-T reaction. 80% of D-T energy is neutrons. These uncharged particles do not interact with the plasma and cannot be used to keep the plasma hot. To keep the process going, you must add high quality energy, which means returning a large part of your electrical energy back to the machine.
Figure 2. Recycling electricity. At each step, a large portion is lost to thermal heat which must be removed from the machine. This sketch shows a reactor with a fusion gain of 20 and zero net power.
You end up recycling almost all your energy, Figure 2. Most of this extra energy ends up in the blanket whence it can in theory be extracted. But what you get back is thermal energy. Suppose super optimistically you can do a fusion gain of 20 which results in a net power out of 5%, and your thermal efficiency is 50%. 95% of your power goes back to the reactor. You get at most half of that back after reconversion to electricity.3 You need a turbine and a generator that is 10 times larger than the net power out. And you need a heat exchanger that is large enough to feed that turbogenerator. A plant that is ten times larger than its net output? That sounds very expensive to me.
For a good fission plant, the turbogenerator needs to be about 6% larger than the net power out. For example, the French EPR has a turbogenerator with a gross rating of 1720 MW. It delivers 1630 MW to the grid. In other words, the net power ratio is of the order of 16, 300 times larger than our fusion machine.
Fusion is setting up to be the next wind/solar, an excuse for not doing fission.
In June, 2023, ITER announced that it will delay by about a year the announcement of how long the latest delay in the schedule will be. This takes procrastination to a new level of sophistication. Is meta-delay a word?
To get that gain, you will need a device whose radius is a little more than 2 meters, (the bottom axis). The blue line at top is wall loading, For our purposes, we can ignore it, even though it is a scary number.
Helion has a concept that avoids the electrical to thermal conversion loss. Their design generates electricity directly in the machine. Assuming they can make this work, Helion faces two big problems:
1) It's a pulsed machine, which means enormous capacitor banks, which must survive many thousand charge/discharge cycles.
2) More fundamentally, Helion uses the isotope helium-3 as fuel. The only way to make helium-3 in the quantities required is conventional fusion.
One of the things that Helion’s seventh generation reactor Polaris is supposed to determine is if the current design will produce enough helion (He3) or if they will need two reactors, one dedicated to making it and one consuming it.
I have back of the envelope figured fusion would need somewhere between a 100x and 1000x gain (2-3 orders of magnitude) higher than 'breakeven' to be economic. This seems to line up with that. Probably at the high end of that range...