The ITER Tokamak is designed to produce 500 MW of thermal output from 330 MW of input power, ignoring electric conversion efficiency.

The heaters require 150 MW (to heat input plasma by 50 MW), the magnets require 80 MW, and the other subsystems require 100 MW (source). How much of this 330 MW input is actually physically required to operate the 500 MW reactor?

For example,

  • Could 150 MW -> 50 MW plasma heating loss be reduced with more efficient heating, or is it nearly optimal / Carnot efficient? (Could input heating be eliminated if a future reactor achieved ignition?)
  • Could any of the magnetic 80 MW be eliminated with better superconductors or a different design, or is this close to the theoretical minimum to maintain these B-fields?
  • Is any of the other 100 MW for auxiliary experimental equipment or non-germane subsystems, or is it all required to couple fusion heat to a heat sink (or electric turbine)?

(If all no, do any/all these inputs scale with the fusion output, e.g. to 5 GW?)

I'm asking about the scientific constraints & theoretical bounds here, not how to construct anything specific or solve any particular problem.


To answer all three of your questions in short: yes.

Future "commercial" fusion plants will hopefully be able to reduce all three sources of power: auxiliary heating, magnets, and other subsystems.

It is impossible to give precise numerical values in MW for how much of this auxiliary power can be eliminated because that would require me to speculate on future technologies which do not exist yet (for example, if we had room temperature superconductors, you wouldn't need any of the cryogenic systems necessary for present superconductors).

Nevertheless, I can comment on some common threads running through various Advanced Tokamak (AT) concepts these days.

  1. (Better) Superconducting magnets - Calculations of the fusion gain, $Q = P_{fusion}/P_{CD}$, find that it is proportional to the magnetic field cubed, i.e. $Q \propto B^3$. Therefore, any new advancements in making higher fields at lower power 'cost' pay off big time when it comes to fusion (in my expression for $Q$, CD = "Current Drive" is the auxiliary power one must put in to drive the large toroidal current in a tokamak).

  2. Efficient Non-inductive Current Drive - AT concepts also like to hope for more efficient ways of coupling power into plasma. To answer your first bullet point: no, our present abilities for power-coupling are not optimal, and this is the subject of active research. Also, to answer your question regarding eliminating heating/current-drive upon ignition: I believe the consensus on this is no. Even in so-called steady-state tokamak operation some amount of power is needed to drive current/heat the plasma. AT concepts typically try to have a large fraction of the toroidal current be self-generated (this is called bootstrap current).

  3. Compact Design - Everyone realizes that ITER is far too expensive (big) to be the gold standard for fusion energy moving forward. So to address your last point about lowering the power-budget for "other systems", I have to imagine that having a smaller machine would reduce power needs. Also, recall that smaller machines are more economically competitive (less expensive, faster build-time).

I would like to refer you to this recent DIII-D Tokamak review as well as this article about the SPARC Tokamak for further reading on more efficient (both in terms of power and cost) paths to fusion.

  • $\begingroup$ I'm skeptical that smaller is cheaper. Fusion energy confinement time scales with $r^3 B$. And realistic power plant would be 500MW–3GW. Lower construction costs can come from economies of scale. I'm curious what the breakdown of the "other systems" 100MW is. Is any of it for non-commercial scientific purposes? $\endgroup$ Nov 16 '20 at 17:20
  • $\begingroup$ @alexchandel , Allow me to clarify, smaller is cheaper in the sense that if you could get the same $Q$ with a smaller machine then you don’t need as many raw materials, hence brining down the cost. I looked at your reference-link for the 100MW number, and it’s a little unclear what this accounts for. In one sentence it seems to imply that this is the “overall consumption of the site” which might account for purely scientific systems, offices, etc. so I too would like to know the breakdown of this number before I can comment on extrapolating it to a commercial reactor. $\endgroup$
    – Quinn
    Nov 16 '20 at 17:48

Not the answer you're looking for? Browse other questions tagged or ask your own question.