Fusion September 2024: Where are we with respect to "engineering break even"?

I'm a physicist, but fusion is not my field. I would appreciate any corrections on any misconceptions exposed in this question.

I understand that in a fusion reactor we supply energy to some sample of matter. This energy induces fusion reactions which release some energy. It is possible for the energy released to exceed the amount of energy put in. We can define $$Q_{sci}$$ as the ratio of the energy released to the energy supplied to the sample. If $$Q_{sci}> 1$$ I've seen this called "scientific break-even". I also think this was realized recently in at least one reactor.

However, the energy required to power the entire apparatus may far exceed the energy supplied to the sample. This is due to many losses in the energy supply chain. For example, the output power of a laser is much less than the wall power consumed by the laser. Also, in the end, the energy released by the fusion reactor will need to be captured and converted into a useful form of energy (almost certainly electrical energy supplied to the power grid). We can define $$Q_{eng}$$ as the ratio of the energy captured and converted to a usable form to the total wall power needed to run the apparatus. If $$Q_{eng}>1$$ we might say the reactor has realized "engineering break even".

The core of this question is: Quantitatively, what are typical values for $$Q_{eng}$$ realized in state-of-the-art fusion reactors?

I've also heard about a concept of "economic break even" which is the point at which the fusion reactor is economically viable. It's not clear to me what the distinction is between engineering break even and economic break even. If you've realized engineering break even then you can buy electricity from the grid, produce more electricity and sell it for more money and make money. Maybe the point is it cost more to run a fusion reactor than just the cost of the energy you buy from the grid. You also need to pay many employees and contractors to maintain the system, you need to pay for materials to build the system, you need to replace parts etc. Perhaps economic break even is the point at which a fusion reactor could realistically make money. So perhaps $$Q_{econ}$$ is the ratio of revenue to expenses for a fusion reactor.

Again, please let me know if I'm getting anything wrong with how I'm framing this question, broadly, or in detail.

• Part of economic break even is recouping the costs of construction. If the profits from selling generated power are small compared to the initial investment, then the expected time-to-break-even is long. If that time is longer than the expected lifetime of the plant, you will never break even fiscally. Commented Sep 5 at 17:59
• You write that it's "not clear to [you] what the distinction is between engineering break even and economic break even," but your description of that distinction seems very good! Commented Sep 5 at 20:48
• Practical fusion is 20 years away, and has been since the 1940s.
– Mark
Commented Sep 5 at 23:28
• In addition to the reasons you quote for $Q_{econ}$ vs. $Q_{eng}$, the cost of the fuel might play a role. For example, what is the cost of the diamond shells ("high density carbon", HDC) needed in inertial confinement fusion? I have no idea... Commented Sep 6 at 10:37

According to my understanding, the current fusion process requires:

• Very elaborately made fuel samples
• Cryogenic cooling for the fuel samples
• The fuel itself which is tritium (of which there is very little on Earth, and has to be produced in fission reactors) or hypothetically Helium 3 (of which there is almost none on Earth)
• Some of the most elaborate and expensive equipment ever made to process the fuel pellets into an energy release

In short, we are a very long way from engineering or economical break-even.

It is worth reflecting that due to the capital investment required, regulatory burden, and comparative cheapness of fossil fuel energy, nuclear fission itself is not exactly at economical break-even, after 80 years and hundreds of reactors built.

• This was a very helpful list for other factors contributing to economic break even. I would love a quantitative estimate of where we are with respect to engineering break even. Commented Sep 5 at 4:25
• Does this answer refer to laser-ignition, or to Tokamak-based designs, or both? Commented Sep 5 at 15:33
• Some tokamak reactors use fuel pellets, but not all. All current designs would need tritium to function. Commented Sep 5 at 23:06

In addition to the other answers, I would like to add that no or very few of the currently operating fusion reactors -- either with magnetic (tokamaks or stellarators) or inertial (laser) confinement fusion -- have any energy-capture system for power generation installed, which further increases the potential difficulty to realize engineering break-even. So there are no experiements which measure the engineering efficiency, simply because there are too many unknowns in what efficiency of such a heat capturing system could be achieved.

With magnetic confinement reactors (which I have some limited experience with), there are various schemes for breeding tritium for the fuel, which might solve the fuel problem. On the other hand, it increases the difficulty of how to capture the heat generated from the fusion reaction and converting that heat energy into useful energy. Next, fuel pellets used in magnetic confinement fusion are much simpler (although still cryogenically frozen hydrogen) than that used in inertial confinement fusion - the latter of which requires incredibly fine-tuned fuel pellets. And the engineering efficiency of the latest laser-fusion results should take into account the fact that only a few percent of the shots are successful at the moment, which further degrades the practical engineering efficiency.

• Right, so this is to say that we have never realized a $Q_{eng}>0$ because we've never even done heat capturing? Commented Sep 5 at 14:17
• @Jagerber48 The mechanisms for capturing heat (i.e. turn water into steam and drive a turbine) are reasonably well tested and understood. The goal of current research is how to better sustain the fusion reaction. So it makes sense to surround the reaction with as much measuring equipment as possible, with just the minimal necessary cooling added. Looking e.g. at the predicted numbers for ITER (~300 MW electrical input, 500 MW thermal fusion output), one could probably just break even using a modern efficient turbine. But that would result in a very expensive scientific dead end.
– mlk
Commented Sep 5 at 14:37
• At my day job we collect the heat output. However, it is so far from mainstream physics that our advisors don't believe our results. Commented Sep 5 at 18:22
• @mlk It's note quite as easy as we understand how to turn heat into work. For one, the inner wall in ITER is water cooled. If this water starts boiling then the cooling capacity decreases dramatically, with the risk of meling the wall. But then if wee keep the water from boiling, then we have very low-grade heat energy which would be thermodynamically unfavorable for generating work. It might be possible with a pressurized system like in some fission reacor, but that would requre dramatical redesign of the walls. Then theres the breeding blanket which has to be inside the water cooling. Commented Sep 6 at 12:25
• @Jagerber48 We have never used any fusion power to create electricity. But it would be theoretically possible to generate some electricity from the cooling-water systems in the existing reactors, albeit with a very low efficiency. Commented Sep 6 at 12:28

Sadly, we are not just very far away from engineering breakeven, but that this measurement would be really difficult to do. For example, the only scientifically accepted $$Q_{\text{sci}}>1$$ result is currently the NIF stuff, which had not just repeated, but suddenly jumped from $$<1$$ to $$1.3$$ to now $$2.36$$. However, that is using the laser output of about $$2.2\,$$MJ; it took about $$400\,$$MJ to get that laser shot. However, that is because they are using gas lasers; just swapping for semiconductor lasers would suddenly lower that to nearly $$11\,$$MJ. And then you need to freeze the DT...

Note that $$Q$$ alone is not really sufficient. We must also have $$\left>1$$

• Ok, so it sounds like with a laser output of 2.2 MJ they generated 5.2 MJ = 2.36 * 2.2 MJ. But you're saying this required 400 MJ. So thats $Q_{eng} \sim 0.013$, right? And then you're proposing thta moving from gas lasers to semiconductor lasers might realize $Q_{eng} \sim 0.47$ which isn't so bad..? Am I interpreting these numbers correctly? Commented Sep 5 at 5:02
• Yes, and so if you try to find a specific estimate, how would you reconcile that there is such a wide range that is coming from using old tech v.s. new tech? Those people in suits might not appreciate the nuance of all these little changes that are so significant to the bottomline, yet they are the ones controlling your budget. I fully understand their need to be careful with how they state things; definitely do not envy their position. Commented Sep 5 at 5:19
• Well, if the recent $Q_{sci}>1$ results are so popular it seems the $Q_{eng}$ results that have been actually acheived should be documented. It's good to know what has technically been acheived so far. And sure, it's good to know what the value would be if certain upgrades are implemented. But it's also important to keep in mind that upgrades are expensive, they usually don't give quite as good technical results as expected, and talk is cheap. Commented Sep 5 at 13:43
• NIF uses Nd:glass lasers (flash lamp pumped), not gas lasers. Part of their proposed upgrade includes going to more efficient pumping. Commented Sep 5 at 14:45
• @Jagerber48 Just to be clear, the facility that managed 2.36 has no plans, and likely will never try to recapture any of the energy produced. Its a weapons research facility. Commented Sep 5 at 18:28

On economic break-even, I can comment, as this is my professional field: as you wrote:

Maybe the point is it cost more to run a fusion reactor than just the cost of the energy you buy from the grid. You also need to pay many employees and contractors to maintain the system, you need to pay for materials to build the system, you need to replace parts etc. Perhaps economic break even is the point at which a fusion reactor could realistically make money.

The cost of building a commercial fusion reactor is unknown, but it will huge (costs of building fission reactors are absurd at the moment, for example, and all the projects being built are more the fruit of political decisions than pure economic-driven ones), and to that you need to add all the operational costs. Usually you need to include an estimation of the dismantling costs when the time comes. And on all that money, you need to give your investors a return on the capital they are spending. This last point, the return on that capital, is which marks the economic break-even. It is only achieved when that margin (expressed as a yearly % on the capital expended) is reached.

The core of this question is: Quantitatively, what are typical values for $$Q_{eng}$$ realized in state-of-the-art fusion reactors?

As far as I know, there are no fusion reactors with the downstream infrastructure that actually generates power. This means any quoted figures for $$Q_{eng}$$ would be a theoretical estimate. The other issue is that both $$Q_{eng}$$ and $$Q_{econ}$$ are highly scale dependent. For example if the heat produced is used to power a steam turbine generator, the efficiency of the steam generator can vary from as little as 10% for very small unit to as much as 40% for a huge industrial steam turbine generator. The scale makes a big difference to the viability and all the current fusion reactors are at the small end of the scale. Actually being able to take the heat and convert it into electrical power will take further research and engineering ingenuity and I don't think we have even started that stage of the development process yet. Most fusion reactors are a batch process. To be useful, they will probably have to design a linear continuous process. I do not know if it is feasible, but ideally we would take the fuel and compress and heat it in something like a straightened out version of the LHC and eventually let the plasma impinge on a water cooled target to absorb the heat to create steam for the turbines. I imagine such a device would have to be at least several if not dozens of kilometers long.

As mentioned above $$Q_{econ}$$ is also highly scale sensitive. For an automated plant, a large plant does not take many more personnel to run it than a small plant so the labour cost per KW output is diminished and there are other savings that always come with scaling up to industrial scales. Exceeding $$Q_{econ}$$ is not as important as exceeding $$Q_{eng}$$, because a power plant that produce electricity on a windless night can fill the gap when solar and wind power are not producing. If a fusion power plant can do that job safely without polluting the environment, then it is useful even if it is running at an economic loss. Its main competition would be lithium battery banks, large gravity batteries and nuclear power stations.

• @PM2Ring Thanks! yes, that is not what I meant to say. fixed that typo now ;-)
– KDP
Commented Sep 6 at 21:28
• Energy conversion efficiency for heat engines does not realistically go above 40%. It is limited by two factors, loss through insulation (scaling effects do apply) and thermodynamic limits (scaling is irrelevant, temperature differential counts). Low temp is ~100°C to keep the steam gaseous, high temp is limited by materials' heat resistance, and that's the main contributor that nails you to at most ~40%. Commented Sep 7 at 18:30
• @toolforger It seems I was misled by the reference I found. I think your figure is closer to reality and I have edited my answer accordingly. This seems a better reference --> power-eng.com/news/new-benchmarks-for-steam-turbine-efficiency/….
– KDP
Commented Sep 7 at 18:49

I've seen this called "scientific break-even". I also think this was realized recently in at least one reactor.

In NIF, yes. But they use a definition of Qsci that you may or may not agree with.

The core of this question is: Quantitatively, what are typical values for 𝑄𝑒𝑛𝑔 realized in state-of-the-art fusion reactors?

For ICF devices like NIF, it takes about 360 MJ of energy to make 2 MJ of laser, which produces 13 MJ of fusion power. We might get 35% of that as electricity if we converted it, so about 5 MJ. So this is very far from Qeng = 1.

It is almost certainly possible to reduce the input energy needed by about a factor of 10 by replacing the flash tubes and glass lasers with diode lasers. This would reduce the input to around 30 MJ, but we're still only getting 5 out. So a couple of orders of magnitude improvements are needed in implosion physics as well.

An alternative concept for ICF is known as "direct drive" which is much more efficient - in NIF most of that 2 MJ goes into heating the target up, not compressing it. Direct drive machines could offer another order of magnitude improvement, at which point we might be able to hit Qeng > 1. However, the physics of direct drive is not nearly as well understood, largely because it is not similar to h-bomb physics and thus receives very little funding.

On the magnetic side, like ITER, the heating of the fuel is much more efficient, with perhaps 30 to 50% of the energy going into heat. The magnets in a production machine would almost certainly be superconducting, and thus use relatively little energy even if you consider the cryo systems. For this reason, an MCF system operating at Qsci > 1 is already much closer to Qeng > 1. You need to consider mostly the conversion to electricity at maybe 35%, and the efficiency of the heating, which leaves you with the figure of merit that a machine at Qsci ~= 10, like ITER hopes for, will be Qeng > 1.

There is another consideration, the "kopeck problem". Consider the output from one shot from NIF, 5 MJ of electricity. In more familiar terms, that's about 1.4 kWh, which here in Toronto runs about 15 cents, tax in. The fuel target to produce that costs tens of thousands of dollars. Even in its far-future predictions, Livermore suggested the fuel targets might reach 25 cents. So from an economic perspective, we are infinity away from fusion.

Things are harder to analyze on the MCF side because there's no obvious "fuel unit" we can consider. Fuel is injected continually and energy is removed continually. The key here is the rate: if we consider the capital needed to build the reactor system, say 30 billion, we can calculate the amount of interest we have to pay on that value on a per-day basis. It is extremely unlikely that that known MCF design will produce enough power to pay for itself, even if the fuel is free.

This last bit is known as "Qeco", for economics. All current designs, even fully developed 50 years out, do not appear to have Qeco > 1.