The greatest problem with fusion power is rarely mentioned and scarcely on the research agenda: capital cost. When I discussed the problem earlier, in “Fusion Power: A New Way to Boil Water”, I hadn’t seen this quietly damning report, which I think is worth quoting:
Issues and R&D needs
for commercial fusion energy
An interim report of the
ARIES technical working groups
July 2008
From the introduction:
The goal of this activity is to provide guidance to the fusion energy sciences community based on industry requirements…
Buried among the discussions of plasma physics, neutron fluxes, and a host of practical engineering concerns, there is a page that briefly notes the “Achilles’ Heel” that makes the rest look like an academic exercise. There is no mention of the problem in the introduction or the conclusions:
From page 22:
Fusion fuel is cheap, but the capital costs are high. This may be the Achilles Heel of economic fusion power. The capital costs must be lowered by significant amounts — an order of magnitude of cost reduction would be highly desirable but probably not attainable. Traditional cost cutting efforts offer marginal improvements and will not be sufficiently effective. Innovative approaches that promise orders of magnitude cost reductions on major items must be aggressively pursued… [This will require] new fabrication and production technologies….
Emphasis added.
Translation: There is no known way to build a remotely economical fusion power plant, even if the fuel is free and the plasma physics works perfectly.
The report speaks of potential, unspecified, orders-of-magnitude reductions in fabrication cost, but what would other technologies look like if evaluated by the same rules?
Advances that would drop the cost of future fusion power machines into a range competitive with current photovoltaic devices are on a scale that would drop the cost of future photovoltaic devices to almost nothing.
(The above all refers to the leading proposal for fusion power, the tokamak approach. I haven’t seen an analysis in similar depth of the competitors.)
As I showed before, here’s the planned ITER reactor, including the high-vacuum chamber and its surrounding high-field superconducting magnets, together with the requisite particle accelerators, power systems, etc.,. Ordinary nuclear reactors are mostly plumbing; this is a fancy physics apparatus, more nearly comparable to the Large Hadron Collider.
For scale, note the person in the blue coat standing at the bottom:
The plasma physics problems are a fascinating distraction from the physics of advanced fabrication. (This would, admittedly, solve the cost problem.)
See also:
- The Physical Basis of High-Throughput Atomically Precise Manufacturing
- Fusion Power: A New Way to Boil Water
{ 19 comments… read them below or add one }
Better title: “Why tokamak fusion won’t provide power”. Other approaches like Bussard’s Polywell design are still quite promising.
I can provide a list of links related to the following series of information
the World Nuclear Association has breakdowns on the economics of existing nuclear fission reactors. the price per Kwh is about 4-7 cents
Cost breakdown nuclear fission
A big factor is interest rates and the time for construction. Longer time to construct means longer time before revenue while carrying the costs of building. Then there is the factor of how long the plant lasts. 60+ year life means more time to spread out the construction costs. A 5 year build with 60 year life gets down to 3.4 cents.
At 45,000 MWd/t burn-up this gives 360,000 kWh electrical per kg, hence fuel cost: 0.71 c/kWh.
Enriched fuel prices could be cut in half by using laser enrichment. GE is making such a plant with scheduled 2012 opening.
Burnup can go up 20 times (various kinds of breeder reactors – like the liquid fluoride reactor that others have told you previously to investigate. 900,000+ MWD/t is possible). If burnup goes up then more KWH per kg and lower cents per KWh.
China is looking to factory mass produce pebble bed reactors with 2 year or less construction times. Walkaway safe from meltdown because they will not exceed 600 MW thermal power per module.
Higher temps mean easier use of “waste heat” for industrial purposes (co-generation). Idaho national labs has achieved 19% burnup of pebbles. Up to 65% burnup is possible with more highly enriched pebbles. Better electricity conversion efficiency is possible with higher temps. China plans to build them by the dozen and share a control center and upgrade to the higher temps, higher burnup, more efficient brayton cycle turbines later. By their 300th reactor module they will be quite advanced.
An inferior fusion reactor can perform transmutation of U238 (waste as it is not fissile like U235 but just fissionable) into plutonium to help 10 regular fission reactors achieve 100% burnup. Transmutation does not require the 20-50 times energy return of a commercial fusion reactor. get to energy breakeven or close to it or just over and you have good enough transmutation. You can also get away with 50% availability (uptime).
So fusion is an alternative to close the nuclear fission fuel cycle.
Leading candidates for fusion –
inertial electrostatic fusion – robert bussard before his death got rid of the magnetic grids to enable 100,000 times lower losses. now has 8 million in US Navy funding. Dr Nebel now leads the project and indicates two years to confirm fusion only commercializability. IEC reactor would be very small.
Tri-alpha energy and Helion energy .Working on field reversed configuration colliding beam fusion. Tri-alpha got funded with over $40 million and has been secretive but may make some announcements this year. Helion built a one third demo and is raising more funds. Note: Helion energy’s reactor would be far smaller than ITER. As tall as a man but about 50 meters long.
General Fusion in Canada with about $30 million. A magnetized target fusion variant.
Dense Plasma Focus fusion- $1.2 million project 2009-2010. Lawrenceville plasma physics. using plasmoid nuclear pinch with about a billion gauss to get high temperatures overcome x-ray losses
Muon fusion continues in Japan. 40% energy return now. They are trying to tweak it to achieve better return.
the big laser fusion projects and ITER tokomak are multi-decade projects. I think ITER is inferior to deep burn fission by itself.
Plasma Physicist Dr. Nicholas Krall said, “We spent $15 billion dollars studying tokamaks and what we learned about them is that they are no damn good.”
http://iecfusiontech.blogspot.com/2007/07/fusion-symposia.html
The above link shows what the utility companies want to pay for power plants. GE’s Vincent Page discusses the issues.
Polywell Fusion has better economic prospects. The US Navy is currently funding the research. As mentioned in the above comment .
Interesting post and comments. I’ll be in France for 4 months later this year to conclude research for an initial history of the ITER project. The politics surrounding the project (as opposed to its scientific merit) will be the focus. What is already of note is how ITER has been castigated by the (small but growing) French anti-nuclear community while regional leaders have lauded it as an engine of economic development.
As several of you have noted, what I say about “fusion power” is really about tokamaks, the dominant approach to fusion today. I’ve been following the evolution of fusion power concepts, including the many alternative approaches, for decades now. All machines that look more-or-less like current tokamaks (stellarators, for example) would have similar capital-cost problems.
Laser-driven inertial confinement schemes are different, but have led to sketches of power plants that again seem highly implausible. The research machines are, however, of scientific interest for studying the properties of various kinds of matter at extreme pressures and temperatures of interest in astrophysics. Plasma fusion machines, unfortunately, are good for little but studying fusion plasmas.
Bussard suggested several fusion-machine concepts, including a scheme for a very different kind of tokamak (with a small, disposable core), and, of course, the entirely different Polywell approach. There’s not much in print about Polywell, at a technical level, but from what I‘ve read, (1) I’d give long odds against the proposition that the scheme actually makes physical, technological sense, and (2) I’m glad to see that it’s being investigated more closely.
The famous Bussard ramjet, by the way, is a non-starter because hydrogen is virtually inert as a fusion fuel, no matter what the temperature and pressure may be. The first reaction on the road to helium produces deuterium (1H + 1H –> 2H + positron + neutrino). This is very slow because it requires the conversion of a proton to a neutron during the nuclear collision (a weak-interaction process). That’s why the Sun has lasted for several billion years, and puts out less power per unit mass than a good compost pile.
Using a fusion machine as part of a fission fuel cycle would be a very challenging way to solve a problem that doesn’t need to be solved. The cost of uranium is such a small component of the cost of fission power that existing, practical schemes for using it more efficiently, such as breeder reactors, haven’t been widely deployed.
@ Patrick — Beware of discussions of “scientific merit”. They’ve been an enormous (and very effective) distraction from the intractable problem of engineering merit. The framing of fusion as a scientific problem is a foundational part of the politics. The program is sold as a way to build power plants, and we know more than enough about the physics to know that the best case result would be useless.
The momentum that has carried fusion research forward in the face of the clear prospect of useless best case performance seems to be, in a sense, a mirror image of the inertia that impedes progress toward molecular manufacturing in the face of the clear prospect of overwhelmingly useful worst case performance. In both cases, the science is clear enough to make the engineering judgment, and in both cases, the application of engineering judgment is delayed by a focus on scientific questions.
Closing the fuel cycle matters if you are looking to scale nuclear fission usage up by one hundred times. Driving down costs by five to ten to fifty times is needed to drive that demand and to transform civilization. Yes, if we are doing things pretty much as we are now then no we do not need to change for several decades or longer.
If I want to get nuclear fission costs to one cent per kwh then the 0.7 cents per kwh for fuel becomes an issue. Better and cheaper enrichment and better burnup help those matters. Fuel fabrication and enrichment are the major parts of making the fuel and there are several ways to make huge differences there either with the fuel or with reactor designs.
Existing reactors can have fuel retrofits with annular or dual cooled fuel for up to 50% more power from the same reactors. (cylinders -more surface area – instead of rods)
Deep burn would also help with the public relations issue of “what about all of the nuclear waste?”. Nuclear waste mostly being unburned fuel. If I want to scale up, I have to solve the public relations issue. Of course this is less of a problem in China and outside the OECD where 75% of the new reactors are likely to go. In China it is straight up engineering questions. So a transition to moderately deeper burn pebble bed from 2020-2035 (starts 2013 but not a major part until 2020) and then a shift to breeders 2030-2050+.
Russia has its 600 MWe breeder that it has been running for that last 30 years in Beloyarsk and is completing a 880 MWe reactor for 2012. Burnup is only about 100-150 Gwd/t. China is buying two fo the 880 MWe breeders and China is making one of their own. India should have five breeders from 2010-2020. The first one is firing up shortly. Russia has 2-3 other kinds of breeder reactors that they are building or will be building. A promising system is the SVBR-100 reactor which would be factory mass produced and would be fast breeder. India, China and Russia all plan to export the reactors that they develop. China is initially focused massive internal build but will get to exporting. Hyperion Power Generation is working on small uranium nitride and then uranium hydride factory mass produced nuclear reactors.
So substantial significant deployment on second generation breeders is coming.
@ Brian Wang — What you say about scale-up and improvements in fuel cycle efficiency makes sense. While the pressures in that direction aren’t overwhelming today, they are substantial and growing, and as the market shows, there are ample reasons for moving to the new reactor technologies.
The effectiveness of these fission technologies is the main reason that I see fusion/fission systems as economically implausible. Fission produces more than enough neutrons with much less cost and complexity.
In connection with fuel supplies, it’s worth noting that thorium, which is more abundant than uranium, can also serve as a nuclear fuel in a breeder cycle. India is pursuing this option, and the top Google search result for [ India thorium reactors ] is in a blog called… yes… Next Big Future!
Costs for fuel for a practical Bussard reactor (if it is practical) can be worked out (roughly) by costing out highly refined Boron 11 used in semiconductor fabrication. Or the depleted Boron waste from the nuclear fission field.
Deuterium prices can be figured from the price of Heavy water.
Dr. Nebel is of the opinion that electricity from a Polywell will come in at 2 cents to 5 cents a KWh. Html check for future reference: ¢
I know a lot of the criticisms here regard to conventional MCF in general, but it is worth noting that ITER is not the only way to build a tokamak, and certainly not how you would go about building a commercial reactor.
In fact none of the big conventional MCF experiments are, and most of what we know about building MCF devices is based on these designs, so it may be a little premature to draw very strong conclusiosn from what we know about the cost of building physics experiments about the costs of building commercial fusion reactors.
Some thought has been put into the economics (and some are favourable, so why stop at one negative report?) but speaking as a scientist working in nuclear fusion I have to agree that it is too science led and there is not enough thought put into commercialisation and engineering as major objectives above and beyond the physics challenges.
Fundamentally, what we have learned about building tokamaks is how to build tokamaks that are good for doing plasma physics research.
A lot of a machine like ITER’s complexity (all that stuff in the computer model shown) comes from the need for scientific diagnostic access. A commercial reactor would presumably be simplified in many areas, even if it were a reworking of ITER.
Moreover, machines like ITER have not been designed to maximise the economics and over time the constraint has arguably moved somewhat in the other direction to avoid nasty surprises: The ITER design is to maximise probability of success of the physics (demonstrating Q > 10, that is more than 10 time thermal power out than is put in, under conservative conditions). In practice this means sticking to what we know (we have had many nasty surprises when scaling machines up in this field) in terms of magnetic geometry, and trying to be conservative. One would obviously not have wanted to gamble on building a spherical tokamak (see below) slightly larger than the current largest tokamak, but which some project would be of similar performance to ITER and at lower cost, only to discover the projections were a bit off and actually it only gets to Q=2 or 3 and doesn’t really give much of a power amplification once all the losses involved in inputs and outputs are factored in.
This attitude is a symptom of the science driven nature of the programme internationally: sometimes it feels like we are studying ways to desiging machines to do better plasma physics, and that a lot of research is done for it’s own sake without the rigid focus of designing a viable fusion reactor.
To get back to the issue of capital costs: issues like aspect ratio can make a big difference. One of the largest capital costs associated in economic studies of fusion power is the volume of the vacuum vessel and volume enclosed by the coils, yet there are reasons to be positive about low aspect ratio tokamaks (spherical tokamaks) which would reduce those big known capital costs, but may of course bring others (higher power densities).
Meanwhile, there are some areas (Plasma facing components for example) where improvement could radicaly simplify the design overall. For example, a low Z PFC able to handle the heat fluxes and neutron dammage and with minimal thermal expansion (I believe diamond films are under consideration) would change the economics in several ways: ELM’s (high heat flux events generated during high confinement mode) and disruptions would be more tollerable so the plasma could be run hotter and in higher confinement; plasma contamination less an issue (carbon being a low Z element); the low thermal expansion means castelation of tiles could be eliminated reducing the area of PFC’s (gaps between castelated tiles must be shadowed) and would reduce tritium retention (a big handling cost); getting rid of berylium as a first wall material reduce handling costs; and if you can repair the film in situ under vacuum conditions (diamond films generated in plasma chemical deposition), then that reduces costs of “casette” divertors and simplifies the vacuum chamber and cryostat… suddenly we are talking about a host of reduced capital costs.
This is of course highly speculative and we should not think in terms of silver bullets that may or may not be around the corner, the point is to illustrate that in terms of reducing the costs of a fusion reactor we should not be thinking in terms of “order of magnitude reductions” of a given component used in ITER as though the best possible economic fusion reactor was represented by an optimised ITER.
Rather we should be thinking in terms of targeting particular aspects of the design that impose constraints on the rest of the system. Perhaps not enough research is done in this area (or at least not followed up on) as the physics led programme is loathe to go into these areas (and lacks the cashto do) while they have yet to demonstrate that they can reach ignition. This is why more facilities like PISCES and a Component Test Facility are necessary, and why a more engineering led section of the program needs to be developed looking at targeting specific constraints that physicists may feel to be acceptable in purely physics terms or no longer even think about as a major area of effort should be targeted at because in economic terms they are acceptable for a research machine, but a major Achilles heel for a commercial reactor.
A long post, my appologies. Perhaps it would be more concise to say:
Before invoking the need for advanced technologies to make a commercial reactor, we should first make sure that it is not possible to build an economic commercial reactor with conventional technologies: experimental MCF devices are conservative in design, bsaed on what is known to work well enough to achieve the desired physics parameters with a high degree of certainty rather than how to achieve the deisred physics parameters at best economic value for a power plant. More thought, effort and risk needs to go into exploring ways to achieve the same performance in a bettre way, rather than simply trying to improve performance by tweaking the design of the experiments, and slapping a steam turbine on the end.
But perhaps not until we actually demonstrate that the physics really works by achieving ignition…
@ Eric – Good point re: science vs engineering. As you might expect, I’m more interested in the politics and sociology behind this mega-project as well as the question of how such projects like this continue to live on -zombie-like. NIF comes to mind as well. I mean, ITER was first pitched c. 1985 when Reagan and Gorbie met in Geneva…and here we are, 25 years later.
I think the tokamak concept has unsolvable imperfections. Nuclear fusion concepts involving electrostatic acceleration are able to use the energy more efficiently. I believe a more practical engineering approach is the aneutronic reactor, magnetic and electrostatic confinement, without neutron fluxes, it can be highly efficient regarding production of electricity.
There is a lot of discussion lately about Liquid Fluoride Thorium Reactors (LFTR) as a realistic policy alternative. Like fusion reactors this fission reactor design promises: 1) Cheap fuel – very large reserves 2) No possibility of catastrophic failures 3) Very little waste production 4) Burning existing waste stockpiles 5) No need for massive containment structures or cooling towers 6) Low potential for abuse through proliferation or terrorism. Most of the promises usually held out by fusion.
Unlike ITER or the exotic alternatives like Polywell, LFTR reactors seem to be mostly “plumbing”. Rather than being tested in 2030 LFTR seems to have been pretty well tested in the 1960′s and 70′s and to have been rejected for power generation more from military considerations than any economic/engineering analysis.
Is a revival of Thorium reactors a practical take on what fusion promised and failed to achieve?
@ Sebastian Tallents — You make good points, but the gap between best-case tokamaks and competing power sources still seems large. These tokamak designs I’ve seen over the years have all or most of the following challenging requirements, all wrapped together in a single, layered structure:
As a way to produce heat to boil water to drive steam turbines, this seems far from practical. I find it hard to imagine closing the gap between this class of system and, for example, the new generation of nuclear reactors, or solar photovoltaics, or solar thermal power systems, all of which are already being deployed. The general problems above apply to a wide range of MCF systems, not just to the tokamak configuration.
By the way, the main problem with pursuing this work, in my view, isn’t the monetary cost: It’s the opportunity cost of consuming the creative potential of excellent physicists and engineers. It is precisely the greatness of the technical achievements in this area that is the problem.
In the fullness of time and with the advent of radically less expensive manufacturing technologies, fusion power sources may well become important. The chief demand, though, may be in regions where the power flux from the big fusion machine at the center of the Solar System is attenuated by distance.
@ Eric:
Re conventional tokamaks I agree*, but this is the point I am making. The focus on the science, which demands comparison on different scales with as much kept similar as possible, and focus on scale rather than design improvements leads to the first main problems you talk about: large magnetic volumes, superconducting coils, huge cryostats and vacuum vessels.
On the other hand, the Spherical Tokamak power plant concept designs suggest you could get to a GW power plant on a machine that was approximately JET sized… (e.g. aspect ratio 1.6, major radius about 3m). For comparison ITER’s major radius is over 6m, DEMO even larger.
http://www.docstoc.com/docs/28084384/Development-of-the-Spherical-Tokamak-Power-Plant/
If the next big programme is building a bigger stellarator, a bigger tokamak, or a tokamak about the size of JET but differently shaped, guess which one the scientists (predominantly plasma physicists) driving the field are going to pick? Plus, in practice, you couldn’t just build a JET sized ST, you have to build the intermediate sized machines to check that power degredation scales as in conventional machines to get the funding: “show us it works in chunks of a 1bn over ten years”.
Ever since people trumpeted Fusion on Zeta, there has been a lot of conservatism in the field, the whole thing has gone down a rabbit hole of proving we can get MCF ignition without regard to whether we can do so usefully. We will probably get ignition, but in the least useful way imaginable.
R.e. Radioactive components is true of a fission device and less of an issue than people make out. Remote handling works ok, and while the activity is high, it does mean you are talking a 50-100 year problem according to colleagues in neturonics, rather than thousands of years. Of course the issues of neutronics and irradiation needs a lot more thought on blanket design, which not much work has been done on as we are all still chasing ever bigger plasmas…
The obsession with fusion being clean means the programme has skipped past hybrid fission fusion machines (actinide burners or fuel breeders) as an intermediate step. We could have done that with JET technology… whether this is better or worse than accelerators and whether you need it if you have a thorium cycle is also a valid question.
We can play the opportunity cost game on everything of course (depending on the assumptions and preferences), but it’s worth noting that the problem isn’t necesarily fusion or MCF conceptually, it’s where the people went into the programme decided to take it. It is possible if things had gone differently that people would now be questioning the need for thorium reactors or other methods for closing the fuel cycle because we have perfectly adequate fission/fusion hybrids that do this job.
But the fundamental point of your post, that ITER doesn’t scale to a good power station (based solely on requiring scalings of the materials rather than a better design) isn’t a fair representation of the field. I think most people working on ITER accept that once they have demonstrated ignition, you need several different DEMOs exploring different concepts.
* (lets leave stellarators aside for the moment, from my perspective they solve a problem that no longer exists while introducing others, thought Stellerator/Tokamak hybrids like the now cancelled NCSX are interesting)
Here is a totally new approach to fusion energy. Go and check it out for yourself at grnenergylabs.com. It cannot be any worse than anything else that’s been proposed plus it can be proven or disproven in 5 short years at a cost of millions not billions/trillions.
Respectfully,
Frank Sanders
Green Energy Labs
It can be a worse proposal if it requires non-conservation of angular momentum. Note that planets do not spiral into the Sun. — Eric
Another problem with fusion is that we don’t have the materials out of which to build the reactors. ITER will experience 10 displacements per atom in the first wall (from neutron collisions) over its life. A commercial reactor will need a material that can withstand at least 75 dpa, and probably 150 dpa or more. There are no materials qualified for this.
In other areas of technology, not having a needed material relegates the concept to slideware. With fusion, apparently it’s considered a minor problem.
Hello, just wanted to tell you, I loved this blog post.
It was funny. Keep on posting!
{ 2 trackbacks }