SoylentNews is people
SoylentNews
from the insert-tab-τ317.25α2'-into-slot-σ902.44β9' dept.
Scientists Start Assembling the World's Largest Nuclear Fusion Experiment:
Fourteen years after receiving the official go-ahead, scientists on Tuesday began assembling a giant machine in southern France designed to demonstrate that nuclear fusion, the process which powers the Sun, can be a safe and viable energy source on Earth.
The groundbreaking multinational experiment, known as the International Thermonuclear Experimental Reactor (ITER), has seen components arrive in the tiny commune of Saint-Paul-les-Durance from production sites worldwide in recent months.
They will now be painstakingly put together to complete what is described by ITER as the "world's largest puzzle".
The experimental plant's goal is to demonstrate that fusion power can be generated sustainably, and safely, on a commercial scale, with initial experiments set to begin in December 2025.
[...] Some 2,300 people are at work on site to put the massive machine together.
"Constructing the machine piece by piece will be like assembling a three-dimensional puzzle on an intricate timeline," said ITER's director general Bernard Bigot.
"Every aspect of project management, systems engineering, risk management and logistics of the machine assembly must perform together with the precision of a Swiss watch," he said, adding: "We have a complicated script to follow over the next few years."
[...] It could reach full power by 2035, but as an experimental project, it is not designed to produce electricity.
If the technology proves feasible, future fusion reactors would be capable of powering two million homes each at an operational cost comparable to those of conventional nuclear reactors, Bigot said.
[...] The ITER project is running five years behind schedule and has seen its initial budget triple to some 20 billion euros (US$23.4 billion).
(Score: 1, Interesting) by Anonymous Coward on Wednesday July 29 2020, @04:17PM (10 children)
If you're interested in interstellar travel, you've probably heard about the idea of a "wait calculation". The idea is that you shouldn't start an interstellar journey while you know that propulsion systems are in rapid development, because there's a strong chance that a ship launched years after yours would pass you.
You paradoxically get there sooner by starting later, and this has nothing to do with the weirdness of time dilation.
I think this notion might apply to fusion research. We keep building elaborate hardware that isn't even going to generate net return. That's just silly. What should we be waiting on?
We should be working on reliable simulations. We have enough data from old reactors. We should find a way to harness the existing and ever-expanding power of supercomputers (which are already useful for other things) to model fusion reactors. It just doesn't seem to make sense to keep building reactor after failed reactor.
Iterating hardware made sense in the days of steam engines and early cars. Each engine or car was cheap enough for a small company or even one man to build. There was no other option. Today we have options, and it appears that physical hardware is always going to require a huge investment unless some AI turns us on to a system that doesn't--but the focus of fusion research isn't in simulation. It's in meat space, it has been for decades, and it keeps producing the same result. What's that definition of insanity again?
(Score: 2) by DannyB on Wednesday July 29 2020, @05:01PM (4 children)
You seem to suggest something like skipping iteration and jumping right to a hash table where the answer just pops out?
Why use a hash table when it is simpler to iterate until you find the answer you want?
To transfer files: right-click on file, pick Copy. Unplug mouse, plug mouse into other computer. Right-click, paste.
(Score: 2) by tangomargarine on Wednesday July 29 2020, @07:01PM (3 children)
Because hashes are O(1) instead of O(n)? [wikipedia.org]
The counterargument to parent's post isn't to say "okay, according to your own analogy you're wrong"; it's pointing out that their analogy is flawed.
Iteration may be "simpler", but when you're talking about millions of rows, hashes are a hell of a lot faster.
"Is that really true?" "I just spent the last hour telling you to think for yourself! Didn't you hear anything I said?"
(Score: 2) by DannyB on Wednesday July 29 2020, @07:18PM (2 children)
(I thought the joke would go without saying that I understand the difference, but I guess not.)
To transfer files: right-click on file, pick Copy. Unplug mouse, plug mouse into other computer. Right-click, paste.
(Score: 2) by tangomargarine on Wednesday July 29 2020, @08:28PM (1 child)
Oh right, you're that poster who somebody else told me "assume everything he says is sarcasm", aren't you.
This is exactly why Poe's Law is a thing.
"Is that really true?" "I just spent the last hour telling you to think for yourself! Didn't you hear anything I said?"
(Score: 1, Offtopic) by DannyB on Wednesday July 29 2020, @09:45PM
You don't get the title of Senior Software Developer unless you know what you're doing.*
10 WRITE CODE
20 GOSUB 10
30 PROFIT
*depending on how your company picks this title, which I had no part in selecting. I just noticed it one day.
To transfer files: right-click on file, pick Copy. Unplug mouse, plug mouse into other computer. Right-click, paste.
(Score: 1, Touché) by Anonymous Coward on Wednesday July 29 2020, @05:18PM
You're a genius! I wonder why no one else has *ever* even considered that?
https://phys.org/news/2018-07-supercomputer-fusion-energy-international-device.html [phys.org]
https://www.hpcwire.com/2019/08/07/supercomputer-powered-ai-tackles-a-key-fusion-energy-challenge/ [hpcwire.com]
https://www.hpcwire.com/off-the-wire/researchers-accelerate-plasma-turbulence-simulations-on-oak-ridge-supercomputers-to-improve-fusion-design-models/ [hpcwire.com]
https://www.nextplatform.com/2018/06/27/cracking-nuclear-fusion-with-supercomputers-and-smart-code/ [nextplatform.com]
https://www.iter.org/of-interest/810 [iter.org]
https://cas.lehigh.edu/sites/cas.cas2.lehigh.edu/files/Halperine_poster.pdf [lehigh.edu]
https://www.tandfonline.com/doi/full/10.1080/10519990500058552?scroll=top&needAccess=true& [tandfonline.com]
https://iopscience.iop.org/article/10.1088/0029-5515/53/1/013007 [iop.org]
https://inis.iaea.org/collection/NCLCollectionStore/_Public/36/115/36115667.pdf [iaea.org]
https://www.comsol.com/blogs/analyzing-a-component-of-the-iter-tokamak-with-simulation/ [comsol.com]
https://aip.scitation.org/doi/10.1063/1.2931037 [scitation.org]
Sigh.
(Score: 1, Interesting) by Anonymous Coward on Wednesday July 29 2020, @06:37PM (1 child)
Experimenting inside a simulation would only work if the following both hold true:
1. You know exactly the laws to encode in the simulation.
2. You have enough computational power to run the simulation at sufficient detail and accuracy and to run it as many times as you need to iterate and find a design that works.
Remember that we're talking about a hugely complex machine with many electrical and mechanical components with at its heart a magnetic field containing a highly compressed and energetic blob of plasma consisting of an incomprehensibly large number of fundamental particles, many of which behave in ways we still do not fully(!) understand.
Since we do not have a perfect theory on how the laws in our universe work together, any simulation we create will by neccesity be an imperfect model of reality. Of course such models can still be useful, depending how accurately they can approximate reality. One reason to do real physical experiments is exactly to get data on how plasma behaves under various conditions, so that it can be integrated in future models.
Another reason for physical experiments is that supporting technologies such as superconductors and metallurgy also need to be developed in tandem. A model showing you the perfect fusion reactor design is not much use if nobody knows how to make the parts for it.
(Score: 2) by PiMuNu on Wednesday July 29 2020, @06:40PM
Not really. Plasma physics, at the microscopic level, is understood very well. Numerical integration can be performed with sufficient accuracy.
(Score: 2) by PiMuNu on Wednesday July 29 2020, @06:38PM
> We should be working on reliable simulations.
The codes exist to simulate this stuff reasonably well, and indeed one can simulate to annihilation - I am sure they already have. Hardware teaches us all of the hard lessons about magnet alignment, management of forces, tuning the containment field, managing instabilities, etc.
(Score: 2) by tangomargarine on Wednesday July 29 2020, @06:56PM
That only works if you *know* there is going to be a breakthrough soon...which such sort of prognostication always sounds dubious, but let's for the sake of argument say it's possible.
Haven't people been working on fusion research saying it's "five years away" for the last several decades? Sounds like we need to keep doing the research, because if everybody else stops, we end up 20 years down the road and those guys who were "just about to invent it" come back and say "oops, I guess that idea we spent all that time on doesn't actually work."
Then you've just taken 20 years off for no reason. Let people continue researching various approaches. There are dumber ways to spend the money anyway.
"Is that really true?" "I just spent the last hour telling you to think for yourself! Didn't you hear anything I said?"