Hacker News Reader

Comment by DennisP | Hacker News Reader

> there are daunting engineering challenges involved in stabilizing a torus where the inboard legs of the magnet coils are subjected to extraordinarily high electromechanical stresses and overturning moments.

True, but MIT (and by extension CFS) has more experience with this than anyone. Their Alcator C-Mod had the strongest magnetic field of any tokamak. I got a tour of it once, and a grad student showed us a metal tie rod, about a meter long. He said they'd calculated that two of them could have held down the Space Shuttle when it was trying to launch. To keep the reactor from flying apart, they needed 38 of them.

> In any case, improving cost-effectiveness is surely a distraction for MCF research when there have been no advances in Q

Except smaller, cheaper reactors mean you can iterate experiments a lot faster.

View on HN

Banana for scale:

It would take a square bar of regular old 250MPa structural steel about 32x32cm to hold the space shuttle, at full throttle, and it's solid rocket boosters down.

Or 38 square bars about 5x5cm.

And 250MPa is just it's yield strength, that type of steel work hardens and has an ultimate tensile strength usually in excess of 350MPa.

There are, of course, much stronger varieties of steel.

Steel is strong

The bar he showed us was quite a bit narrower than 32x32cm, so if there are much stronger varieties, that pretty well fits.

For sure, you'd likely have heard of chrom-molly steels.

Got example, 4140 is a 1% chromium - molybdenum medium hardenability general purpose high tensile steel - generally supplied hardened and tempered in the tensile range of 850 - 1000 Mpa.

Three to four times greater tensile strength, so two 127mm (5") diameter rods would do it at 1000 MPa.

Google says it's Inconel,

- "The in-plane loads are carried from the covers to the cylinder by 96 INCONEL® 718 draw bars with a yield strength of 1 GPa." [0]

- "The wedged conductors are CD107 copper, bonded to flat Inconel 718 plates, with a 4:1 copper/reinforcement ratio over the central column area. The yield strength of the copper is 310 MPa, up to 100 °C, while the yield strength of the Inconel plates is 1130-1300 MPa. The peak Tresca stress on the Inconel at the inside radius is calculated to be 580 MPa, while that in the copper is 310 MPa." [1]

[0] https://www-internal.psfc.mit.edu/research/alcator/data/fst_...

[1] https://dspace.mit.edu/bitstream/handle/1721.1/94967/87ja040...

Speaking of fruit, I’m interpreting this as an apples and oranges situation. If we’re talking about trust of the engine that’s one thing, but holding back a launching rocket is different. That is the absolute worst case for the rocket (and best case for the rod) due to the weight of the heavy fueled up rocket fighting the trust along with the rod

That's a good point, I was calculating based on max thrust only. If we take in to account all the weight at take off, it'd be a much smaller section.

Depending on how long the steel has to hold the rocket down :)

Instead of a ring-shaped tokamak with magnetic confinement, the thought came: why not a 1-dimensional Möbius strip?

After a brief bit of web searching, it turns out that the design exists, and it's called a Stellarator:

https://en.wikipedia.org/wiki/Stellarator

And if you're ever thinking about building one at home, just try not to cause an accident. What's more dangerous than playing with fire? Playing with smoke detectors.

A desktop Z-pinch device is actually fairly doable to build at home, it's a (very) simplified version of the stellarator.

are there any builds you could link? I've seen people make Farnsworth furious on hackaday etc. and I'd love to see this.

There are quite a few Z-pinch builds on YouTube I think

It just a vortex.

Both stellarators and tokamaks appear as Möbius strips to confined charged particles. The key difference is that the toroidal transform is applied via induced toroidal current in a tokamak whereas it's inherent in the magnetic field geometry of a stellarator. We're in the middle of the golden age of stellarator geometry research and optimization. It's a very interesting problem with large payoffs for modest investment.

> two of them could have held down the Space Shuttle when it was trying to launch. To keep the reactor from flying apart, they needed 38

And all that for a reactor that was unable to break even in energy terms, let alone ignite and break even economically.

It perfectly illustrates the folly of magnetic confinement: to hold the plasma you need fields so strong and equipment so massive and expensive that the capital expenses get out of control, fully negating any advantage fusion might have over fission. We already know how to make capital intensive energy - the fueling costs are negligible for a fission reactor, yet those largely failed in the market. An even more expensive and capital intensive solution is dead in the water for economic reasons regardless of any theoretical merits.

Mechanical strain has never been anywhere near a limiting factor on the engineering of MCF machines.

With high temperature superconductors, mechanical strain is in fact now the limiting factor, at least for the magnets.

I must have missed the paper highlighting this fact. Could you share it?

If you look at the 2014 ARC paper, you will discover that 60% of the mass of the reactor is the stainless steel structure for resisting the JxB forces on the magnets.

I never said it was. But a device that needs multiple square feet of high tensile steel to hold together will have immense costs because it also includes gigantic superconducting magnets, vats of cryo-coolant and myriad of custom, unique parts made of exotic materials for exorbitant money.

The billion price tags of tokamak experiments is not a scientific oddity, it's a fundamental economic issue with the approach that will carry on to any production plant, that as far as we can conjuncture will need to be even larger and more expensive.

Nonsense. That's all gravy. The cost difference between experiments and reactors is scale. ITER is a good example to point to. If reactors needed to be the size of ITER mankind could never deploy them at scale because we simply don't have the resources. If ITER was 5x smaller: it's a different story.

Since ITER's volumetric power density is 400x worse than an existing commercial PWR's reactor (+ its pressure vessel), we'd need quite a bit more than a 5x reduction in volume. Even a 5x reduction in linear dimensions (likely impossible at the same power output, due to wall loading 25x higher) may not be enough.

> mankind could never deploy them at scale because we simply don't have the resources.

I don't understand what your objection exactly is. That's just a way to rephrase my statement: magnetic confinement, as it stands, is too capital intensive to ever break even financially, regardless of the experimental success ITER/DEMO might have.

I agree if confinement field tops out at 12 T on coil / 6 T on axis. Double it and the story looks very different.

Some light reading on this:

https://www.cambridge.org/core/journals/journal-of-plasma-ph...

Reply on news.ycombinator.com