Why fusion will never happen October 26, 2012Posted by Maury Markowitz in nuclear.
I like fusion, really. I’ve written some measurable amount of the Wikipedia’s entire collection of articles on the topic of fusion power (including most of this one). I’ve talked to some of luminaries that work in the field, they’re great people. I love the technology and the physics behind it.
But fusion as a power source is never going to happen. Not because it can’t, because it won’t. When you read some of the reasons, you might go so far as to say it shouldn’t.
I didn’t want to have to write this article. I mean it. But glowing fluff articles are pretty much all you can find on the topic. There’s lots of technical papers on it out there, but they’re hundreds of pages long and filled with math. The ‘net seems to lack something in between those extremes.
As nasty as it might seem, a little dose of reality is clearly needed. So here we go…
A little chemistry
Chemistry is the study of electrons in atoms, and how they interact. Very simply, some atoms hold onto their electrons very strongly, and most others less so. If you take some of those “less so” atoms (everything but the noble gasses) and put them together, they might stick together and form a molecule. If the molecule is more stable than the individual atoms, energy will be released. Normally this comes out as heat.
Consider coal. A major component of coal is C2 – that is, two carbon atoms holding onto each other. You can break them apart by applying a little force, 83 kcal/mole in chemistry-speak. How do we supply that energy? The easy way is heating it up, in this case to about 700 degrees C. If you then mix that with oxygen, a new molecule becomes possible, CO2, which is more stable that the original C2 or O2 you feed into it. That reaction, forming the CO2, gives off 354 kcal/mole. That’s around 2000 degrees C.
Now here’s the cool thing about this… 2000 is higher than 700, which means that once you get it going, the heat being given off by one bit of coal burning is enough to light the piece of coal beside it. We call this condition “ignition”. Ignition makes life easy.
These sorts of reactions can occur in any “direction”, you can make more complex molecules, or simpler ones. All that you need is to balance the energy – either have reactions that give off net energy, or, more rarely, supply external energy to make up for any losses.
Even less nuclear physics
Nuclear reactions are pretty similar in the basic concept to chemical reactions, but this time you’re considering the nucleus of the atoms rather than their electrons. These are held together much more strongly than the electrons, so the energy scales are way higher. That means we need more energy to start a reaction, and get a lot more out when it’s done.
In the case of a conventional nuclear plant we use atoms that are inherently unstable – if coal can be thought of as somewhat unstable, uranium is downright crazy. Left to its own devices, uranium will periodically “decay” into other atoms, those will then also decay, and this continues until you end up with lead. When this happens it gives off lots of heat, although it’s spread out over many, many years.
But another thing that happens during this reaction is that neutrons, one of the parts of the nucleus, go flying off. If one of them hits another uranium nucleus under just the right conditions, they cause that nucleus to decay right now. This is the basis of the “chain reaction” you’ve heard of. Nuclear reactors are basically piles of uranium that are carefully positioned and controlled so that one of the neutrons coming out of a reaction causes another one to go. That perfect balance is called “criticality”, although it’s pretty much identical in concept to ignition.
Just like the chemical case, nuclear reactions can also occur in “both directions”, but in this case we don’t produce more or less complex molecules, we produce heavier or lighter elements. I won’t get into the physics here, but basically heavy elements want to break apart, and lighter elements want to clump together. It all has to do with the number of neutrons and protons in the nucleus. The breaking apart is fission, which we use all the time to generate power. The clumping together is called fusion, and that’s why we’re here today.
If the basic rule is lighter elements want to clump together, then you might guess that hydrogen is the best fusion fuel. Almost correct. The almost is because of those neutrons – hydrogen will stick together much more easily if it already has a few extra neutrons. So the best fusion fuel is not hydrogen, but hydrogen’s two isotopes, Deuterium (D) and Tritium (T). The former can be extracted from water, the later we can make in the reactor itself.
So we saw earlier that as long as the conditions are right, we can ignite the fuel and keep the reaction going on its own. In the case of coal you just have to heat it up enough. In the case of uranium, this “energy barrier” is so low that it happens naturally around us, to the point that several nuclear reactors self-assembled billions of years ago.
Fusion reactions can also ignite. They require 0.13 MeV to get going, and then release 17.6 MeV when they react. That gives us lots of extra energy to keep things going. All good, right? Well, not so much… here, let me convert for you:
0.13 MeV is 400 million degrees C
17.6 MeV is well into the billions
See the problem? Keeping burning coal in a container at 2000 C is hard enough, but we can do it. 400 million C? Enjoy!
To date, the only way we’ve been able to do it is to take an atomic bomb, focus the energy coming out of it, and use that to squeeze the fusion system together for a few microseconds. We call that an H-bomb. We haven’t figured out how to contain one of those, under several meanings of that term. Not that we haven’t tried. And tried.
Now it’s bad enough that we need to bottle up something that’s in the millions of degrees, but it turns out its a lot harder than even that.
The whole idea of fusion is to heat up the fuel until it’s all moving so fast that they bump into each other with more than 0.13 MeV. The good news is that when you heat something up, some of the particles inside are going fast, and some slow – remember, the definition of temperature is the “average kinetic energy per unit volume”. In this case it’s that’s average that’s key. That means that even at lower temperatures, some of the particles, a few per million say, will bump into each other hard enough to fuse. And when that happens, all that new energy comes out.
So in theory you can sustain a fusion system at temperatures that are much lower than the theoretical ignition temperature. If we can keep that energy inside the system, then it will heat up the surrounding fuel, and so on. But how much? And for how long?
Back in the 1950s a very smart guy named John Lawson figured it all out. There’s no single answer, instead, the Lawson criterion suggests a whole range of answers. The best point along the resulting curve is at about 10 million degrees at about atmospheric pressure and keeping it all there for 10 seconds.
Now imagine trying to bottle up something that’s 10 million degrees for 10 seconds. You might imagine that it’s not all that easy. And it’s not, we’ve been trying for 60 years now, and still haven’t reached ignition yet. But it certainly isn’t impossible, we have all sorts of ideas on how we could do it, and some of those have progressed a long way on the road to actually doing it.
So it appears that it’s largely an engineering problem.
An engineering problem
Submarines are the sharp end of any major naval force today. They hide underwater and then strike without warning, using weapons that can break a ship in half, or reduce entire cities to burning rubble.
All we have to do is put them out in the open ocean where they can hide, and then they’re basically invulnerable. But of course they need to be serviced and rotate their crews, so we’ve always got some, often the majority of them, close to shore. To keep enough of them out in the ocean to be safe, we need to build lots of extras. The UK, for instance, has four ballistic missile submarines so that they can have one at sea at any time.
Or we could build a fast submarine. Then they could stay in port whenever they wanted, and run out to sea when they were needed. The definition of “fast” keeps changing though. Back in the 50′s, fast was “a couple of hours”, because it would take a good chunk of a day for the bombers to get to you. But by the 1970s, fast was “about 15 minutes”, which is how long a submarine-launched missile takes to get to you. That’s a whole other magnitude of problem.
What could move them that fast? A large cargo aircraft could. So what we could do is build cargo aircraft, load the submarines onto them, and leave them parked on a runway. Or better yet, we could build strap-on wings and engines to the sub itself. Just park the resulting subma-plane at the ends of the runways, ready to go. When it’s go-time, they seal the hatches, fire up the 50 engines, and off they go at hundreds of miles an hour. After 10 minutes they’re way off shore, and dive into the ocean and disappear. It would have to be big. Really big. Bigger than Trudeau’s flying oil tanker project.
If you’re in the Navy, the flying submarine sounds like a really good idea, and you’d tell everyone about how great it would be. It will take a while to develop and we’d have lots of prototypes that don’t really work, but you’d confidently say that with enough development we’re sure we can do it.!
If you’re in the Air Force, or anywhere outside the Navy really, it’s a moronic idea because there’s a perfectly good replacement that already exists. You just put the missiles in existing cargo aircraft and get 100% of the same end result without spending a dime.
But make no mistake, we could do it… it’s just an engineering problem.
And there you have it
A certain minority of physicists have been telling everyone for decades that fusion is going to be the next big power source. Sure, it’s going to be really hard to get it to work, but… it’s just an engineering problem.
And boy do we need fusion! It’s unlimited, safe, clean, and cheap. It’s the perfect source of power.
Except it’s not, well, any of those things. Some of those boasts are widely inaccurate. And even if we do get it to work, it’s not clear we’d want to.
Fusion is the flying submarine of energy.
First, the unlimited part
Fusion reactors, at least in the foreseeable future, would run on D and T for the reasons outlined above. As fusion proponents are quick to point out, the stuff is just lying around in the oceans, ready for us to scoop it out. Except for one thing. Only 1 in 20 million water molecules is D2O. So “scoop it out” is more like “use a really expensive reaction” known as the Girdler sulphide process, which makes the stuff cost $300 a litre.
Now admittedly, $300 a litre isn’t really all that much, considering all the energy that’s theoretically in there. But that only gets you to “reactor grade”. And we’re not talking fusion reactor here, we’re talking fission reactor, for things like CANDU. What, you want something pure enough for a fusion reactor? Oh well, that’s so expensive we don’t even know how much it costs, the techniques needed for large scale use are still under development.
But that’s fine, the fuel is really out there, and if power gets expensive enough, any price will be low enough. Right?
But that’s not the end of the story… D is out there in the ocean, but the T isn’t. Tritium decays rapidly, and even if you make a bunch of the stuff, a couple of years later and it’s gone.
So what you do instead is “breed” the tritium. You surround the reactor with liquid lithium metal, and when the neutrons from the fusion reactions hit the lithium, out comes tritium. This basic reaction is also used in H-bombs to generate most of the tritium they burn.
Now that’s great, but it kinda just moves the problem from A to B. We can ”generate” enough of the stuff, but that pre-supposes we have enough to get going in the first place. And in order to generate it, we need lithium, and although it’s literally lying around for the taking, it’s not exactly cheap. You know, like $4,200-4,500 a tonne.
So really, how much of this stuff do we need to get things going?
The ITER experiment, the largest conventional fusion reactor (being) built, will run on 20 kg of tritium. This is the entire supply of tritium generated by the CANDU fleet, one of the largest sources in the world. Make sure you understand that – a small testbed reactor will use up pretty much all of the tritium in the world. It’s estimated that a single 1 GW commercial reactor would burn that up in only two months of operation.
So basically if we end up figuring out how to make a commercial reactor, we can’t really start it up. Even if we collect enough T for it, we can only make one. Now that one reactor will start breeding more tritium for the next one, but that will take a couple of years to generate enough to fill another reactor. Eventually you’ll get a couple going and then you start to get the I-told-two-friends effect, but it will take decades.
How many decades? Well no one really knows. What we do know is that there’s an inherent catch-22 involved… if we try to get more T out of the system more quickly, we run out of T to run that reactor. If we take it more slowly, we lose more due to decay. It’s really annoying. Now we have a bunch of guesses on how to speed all of this up, but we haven’t even tried a single one of them. Honestly, we have no idea how to really make this work.
This certainly isn’t an impossible problem to solve… it’s just engineering.
Second, the clean part
So we’re going to use those neutrons to generate tritium, feed that T back into the reactor to make more neutrons, and so on. So far so good.
But that lithium blanket? It’s liquid. So we have to have something between it and the plasma. Normally that’s a steel tube. The neutrons have to get through that steel into the lithium. Most will, some won’t. And the ones that don’t will get stuck in the steel, or the surrounding building, or anything else nearby. And when that happens, all sorts of nasty things result. For one, heavier materials like steel will pick up some of the neutrons and become radioactive. That’s bad, but what’s worse is that all those neutrons smacking into them causes the materials to break down, physically.
What that means is that you have to replace the liner materials really often, on the order of months or years. One design used this as a virtue, and planned to replace the entire reactor core every 30 days or so. But no matter how you slice it, you’re going to have a really bad duty cycle because the plant is shut down all the time, that’s does terrible things to your economics. Worse, a lot of that material is highly radioactive, which makes this whole process more dangerous and time consuming.
There is some good news here… generally when something is highly radioactive, that implies that it will “burn out” quickly. Unlike spent nuclear fuel from a fission reactor, which is mildly dangerous for thousands of years, a fusion core is really nasty, but will only remain that way for a couple of decades. That’s actually a smaller problem, I’m pretty sure we can figure out how to keep that in a pool of water for that long. Because nothing could happen to a big pool of water, right?
So we still have a problem with nuclear waste. It’s not the same sort of problem as with fission reactors, but it’s a problem nonetheless. And since there’s lots of metal in these cores, and you have to replace them all the time, the scale of the problem is considerable. There’s no way you could possibly call this “clean”.
Numba three, safe
So now we have two very real-world examples of what goes wrong when a fission reactor gets really out of control. It’s not pretty.
The pile is a construct designed to meet certain geometric criterion to get the right “neutron economy”. Unfortunately, that geometry requires far more fuel than you’d want ideally, there’s months and months of fuel in there. That’s fine when everything’s running right, and you can even see an advantage by putting in extra fuel and not having to refuel the thing all the time.
The downside is that once it ignites, goes supercritical, you have months of fuel to burn off. And that fuel gives off tremendously radioactive waste products that tend to get into the environment when these sorts of accidents occur. It takes years to clean these things up.
The big advantage to fusion reactors is that they contain relatively tiny amounts of fuel, a few seconds worth, or in some designs, microseconds. The amount of energy in the fuel isn’t enough to total the reactor even if it does run away from the controllers. In a few seconds it will burn out, maybe with some damage to the reactor, but nothing catastrophic.
In fact, the most dangerous failures are mechanical – most forms of fusion reactor have very powerful magnets as a basic part of the design, and they operate under tremendous stress. The failure modes for these can be nasty, throwing chunks of metal about. But that’s no worse than any other mechanical problem on systems of this sort of size, like a turbine failure. And even in that case a small amount of fuel will leak, with only a little bit of tritium, and some radioactive components that are all inside the building. It’s a totally different order of magnitude compared to a fission reactor failure.
Except for one thing…
Remember all that about the breeding of fresh tritium? All of that tritium is stuck in the liquid lithium. Which is highly flammable.
That’s right, its a highly flammable, boiling, highly radioactive liquid metal. Fun stuff!
If that’s not enough, it’s also highly reactive chemically. In fact, it can burst into flame when it comes into contact with water.
That’s right, water.
So the real failure mode is any sort of accident where the lithium gets out and catches on fire. In this case enormous amounts of ash will be released, mixed with several kilograms of tritium.
And the problem with tritium is that it’s basically hydrogen, which is lighter than air and tends to leak out of any container. And because it’s hydrogen, the thing it really likes to do is burn with the oxygen in the air to make hot “tritiated water”.
And what does hot water do in the air? Cools off and falls as rain.
That couldn’t possibly be bad.
So how much of this are we talking about? Well let’s compare it to some yardsticks:
Three Mile Island released 20 Ci of 131I.
Chernobyl released 700,000 Ci of 131I
Above-ground nuclear testing in Nevada released about 150,000,000 Ci of 131I.
Now this yardstick is based on 131I, which is an efficient carcinogen. Tritium isn’t, it does its damage in other ways, but it’s equally deadly and much more fast acting. So, some numbers…
T’s reactivity is about 9.8 kCi/g, so a 10 kg burnup that efficiently turned into T2O would release 1,000,000 Ci. That’s more radioactivity than Chernobyl.
Does that sound safe to you?
But really, this is not a big issue. All we need here is a really good containment building with great fire suppression systems. It’s just an engineering problem.
But wait, there’s more.
Tritium is hydrogen. Hydrogen, as I mentioned earlier, leaks out of everything. Not a little.
Back in 1973 the AAAS estimated that a single commercial reactor would leak about 20 Ci of T every year even with the best possible containment systems and no abnormal operation. That’s a Three Mile Island at every plant, every year.
That’s not a huge amount of radioactivity, but if you consider any sort of wholesale “fusion economy”, you’re going to have a serious background level being re-created all the time. And since it has a half life of about a decade, the amount of it in the air will be much higher than the yearly amount, say 10 times. So if we build 500 of these world-wide, that’s 100,000 Ci of T floating around all the time.
That’s not a small amount of radiation. It’s 1/10th of a Chernobyl. Forever.
Update: I found some more numbers of this here… ITER is expected to release about 10 µSv of tritium a year. It’s limited to 100 µSv in the expected worse case. A Sv is a unit of radiation that also takes biological effectiveness into account.For comparison, you’ll get about 2000 µSv a year from the background around you. But now imagine if we replaced all of the fission reactors in the world with fusion ones, they were the same size as ITER, and they all leaked the same amount. That’s about 430 plants, so maybe 4,000 to 40,000 µSv a year, building up for, say, 10 years. That’s spread out over the world, but it’s still nothing to laugh at.
Of course that was in 1973. I’m sure things have improved since then. After all, it’s just an engineering problem.
Wait, did you think we were done? Sorry. One of the other “design issues” is that the millions-of-degrees fuel periodically escapes the system and hits the walls, burning them off and dumping the ash into the fuel. If the walls are some sort of heavy element, like steel, the atoms mix with the fuel and cool it off. Bad.
So what they do is line the inside of the steel container with something much lighter, normally beryllium. Beryllium is a toxic metal. It’s so nasty that a competing design that doesn’t use it specifically points out how bad it is, according to them it’s a “a substance that has health and safety, economic, supply chain, and public perception challenges.” Sigh.
Drum roll please… it’s time for cheap!
So ok, we have a bunch of problems, but we humans love to solve problems. In the grand scheme of things, fusion is safer than fission, and in the grand scheme of things, fission’s been pretty safe over all. Assuming we could solve all of the actual problems getting it to work, fusion might indeed overcome all of the issues mentioned above.
But the question is why? Why go through all of this effort? Is it worth it? Well according to supporters, that one is a no-brainer. The price of power keeps going up, and sooner or later the price of all this engineering will become worth it.
So let’s just consider at what a fusion plant would look like. On one side it’s basically identical to any existing “heat engine” power plant, like coal. These are normally built in two halves, the “burner” (or whatever is making the heat) on one side, and the “turbine hall” on the other where the power is extracted in steam turbines. If you look at pictures of power plants, you’ll almost aways see that in the plant layout, normally a tall side with the burners and a lower side with the turbines. The oddball is the natural gas plant, which burns the fuel right inside the the turbine.
In the case of a coal plant, the heat engine is about the same complexity as the turbine side. Coal plants are the cheapest form of power plant to build, about $2/Watt – meaning that if you want to build a 1GW power plant, you should expect to pay about $2 billion for it.
For a fission reactor, the the turbine hall is pretty much unchanged. But the burner side, the reactors, are much more complex. This drives up the cost of plant to about $5 to $9/Watt, depending on who you ask. Since the price of electricity is basically the price of the plant divided by how much power it produces, nuclear power costs more than coal. Yes, even when you factor in the fuel costs.
So what about fusion? Well the easier answer is “we don’t know”. But we can make some estimates. What we do know is that any sort of baseline fusion plant with be fantastically complicated, orders of magnitude more complex than a fission design. It is filled with incredibly complex machinery for fuelling and tritium extraction, all sorts of heating systems, superconducting magnets and all their support machinery, the heat extraction system which has to be built in two separate parts (some of the heat goes into the lithium, some doesn’t) and ridiculously complex control systems. And then on top of that, the energy they produce is so diffuse, you need huge plants; even the cost of construction will be enormous.
Any one of these issues would make it as expensive as a fission reactor. Together, they’ll cost orders of magnitude more than any existing power source. And then there’s the fuel on top of that, and the limited lifetime of the reactors. If you plug in any reasonable assumptions for any of these costs into a basic LCoE calculation, you’re going to come out with electricity that’s way more expensive than anything we currently use.
In 1994, the US Electric Power Research Institute, a joint R&D group run by the of the U.S. electrical companies wrote a lengthy report on the issue called “Criteria for practical fusion power systems“. In it they describe the variety of problems that fusion power would have if it were to be adopted – not technical issues, basic issues like the paperwork required to build a plant, and even public perception issues.
Although they draw no conclusion directly, the paper states that in order for anyone to even consider building one, then “To compensate for the higher economic risks associated with new technologies, fusion plants must have lower lifecycle costs than competing technologies available at the time of commercialization.”
So a fusion plant cost less than what we have now? No. In 2012, Robert Hirsch, who used to run the U.S. fusion program, stated it clearly:
First, we have to recognize that practical fusion power must measure up to or be superior to the competition in the electric power industry. Second, it is virtually certain that tokamak fusion as represented by ITER will not be practical.
Hirsch then goes on to propose a different form of fusion power as the solution. That’s not too surprising – when you’ve been working on a hammer for 40 years, everything starts looking like a nail. Simply put, there’s zero evidence that his suggested approach will ever work, let alone solve all the problems he so clearly presents. That’s no slight on Hirsch’s plan, the same is true for every proposed approach to fusion. None of them work. At all.
For fusion to happen, we have to overcome enormous technical challenges, build huge power plants to harness it, build a huge infrastructure to support it, and pay for it all. We don’t know if we can really afford to do any of these things.
And why bother?
The basic argument for fusion from the start was that electricity prices will keep rising, and all of our existing sources will eventually run out. And those statements were probably true during the era when then were posited – in the 1950s when most of our power came from hydro or coal, and the usage rates were growing exponentially.
But today? Here’s the real problem for fusion, PV is already less expensive than fusion could be, and wind is even cheaper. Neither of those even uses fuel, and both could generate more power than that world uses.
No, I’m not saying we should use wind and PV for all our power. But what I am saying is that the very niche that fusion would have attacked is the very one we can solve with existing technologies if we want to. And unlike fusion, they actually work. Today.
Quite simply, we need to stop working on this. I fully support a pure research program for new approaches to fusion, but any money spent on the baseline approach like ITER is simply throwing good money after bad. Fusion is not the droid we’re looking for.
Here’s some light reading on the topic
- Allen L Hammond, William D Metz, and Thomas H Maugh II, ‘Energy and the Future’ Washington DC, American Association for the Advancement of Science, 1973.
- Lawrence E Lidsky, ‘The Trouble With Fusion’, Technology Review Vol. 86 October, 1983. Pages 32-44.
- D. McMorrow, ‘Tritium‘, JASON, The MITRE Corporation, November 2011
A reader on Sladshdot noted that the article above refers entirely to problems with DT fusion. I don’t believe that’s really true – the economic arguments remain no matter what the fuel is – but generally many of the nastier problems above are due to the neutrons and their handling. There are fusion cycles that use fuels that don’t produce (as many) neutrons, and therefore don’t suffer from exactly these problems.
Ok, so let’s deal with that. To start with, even the simplest of these “aneutronic” cycles are orders of magnitude more difficult to accomplish than DT. We really don’t have any idea how to build a machine that can burn these fuels. If that weren’t enough, they produce less power output. So basically there doesn’t seem to be any possibility that these could be made to work economically in any future where anyone reading this is still alive to see it.
The topic keeps coming up, however, because proponents of a number of fringy fusion machines claim they can do it – in spite of failing to get even close to working even with DT. A common example is the Polywell, a near mythical device that would almost certainly work if they could just get a bit more money, but it’s being kept down by The Man. Actually, it’s being kept down by plasma physics, but in spite of this, supporters keep grasping at it as tho holy grail of fusion. Similar machines like focus fusion and a host of others are similarily fawned over. Keep the Dream Alive!
What’s more astounding is that the favoured aneutronic fuel is Helium-3, which is not even available on the Earth. Supporters propose building moon bases to mine it. The obvious ridiculousness of this concept is out of this world. Nevertheless, it’s worth looking at the math yourself.