optimal geometry of rods to produce desired plasmoids?

[cccccttttt]

——-

Years of tweaking and computer simulations have brought the Focus Fusion device very close to Boron ignition.

So this question is a detriment to the immediate effort, but will surely be asked by future engineers:

What is the optimal geometry of the rods to form a plasmoid with the most desirable features for fusion?

Configurations yet untested could produce plasmoids using less energy, or that form quicker, or that collapse with a more useful effect.

Three cheers for the current prototype, but it may look quite different in several decades.

ct

[jamesr]

I suspect however good the theories and computer models get, in the end trying out lots of ideas through mutation & selection will be a faster way of improving the design.

Something like this: http://www.youtube.com/watch?v=END3r3ehcDw

[cccccttttt]

Agree.

Imitating natural selection has been used in computer programming to find better algorithms.

Interesting how the technique might generate new designs for an optimal Focus Fusion device.

In the meantime nature has had a long time to produce plasmoids and there must be tricks

to be found just by careful observation.

ct

[delt0r]

I now work in Evolutionary biology (was physics). I assure you its not as good as you think. In fact if you understand whats going on, there are often simple things you can change for improvements that a mutation selection thing never gets. For example solar cells are far more efficient that photosynthesis.

First seek to understand….

[cccccttttt]

You make a good point.

However, an idea born in a US research lab (human insight) is often refined through Japanese engineering

(natural selection) into a superior product.

Focus Fusion researchers are struggling to build the first boron burning prototype, but later will

come years of tweeking to find the optimal structure.

ct

[Henning]

Simulation isn’t as advanced as you may think. There are particle-in-cell simulations but with a density much too low for our purposes. Then there is a parametric calculation that tells you what you might expect with different diameters and pressures and so on.

As third solution there is the LPP simulation, which simulates the plasma as a fluid (with extra quirks), because you cannot handle individual particles anymore at this density. It only simulates a single plasma filament. And it isn’t finished yet.

None of them include the shape of the DPF, except being a cylinder, or a sub-part of it.

[cccccttttt]

Thanks for the reality check.

Easy for an outside observer to fall into the fantasy trap.

Cars once modeled in clay are now designed digitally.

Nuclear bombs are now tested using finite element simulations.

Wind tunnels are replaced by supercomputers.

We need FF now by whatever methods are available.

Keep up the good work.

ct

[jamesr]

Henning wrote: Simulation isn’t as advanced as you may think. There are particle-in-cell simulations but with a density much too low for our purposes. Then there is a parametric calculation that tells you what you might expect with different diameters and pressures and so on.

As third solution there is the LPP simulation, which simulates the plasma as a fluid (with extra quirks), because you cannot handle individual particles anymore at this density. It only simulates a single plasma filament. And it isn’t finished yet.

None of them include the shape of the DPF, except being a cylinder, or a sub-part of it.

There are plenty of PIC codes that can handle the modest densities of a DPF plasmoid (ie around solid density) Inertial confinement simulations have to cope with 1000 times higher.

No one simulation would be able to cope with the full range of scales involved. I would think at least three simulations would be needed, each feeding into the next. First a cylindrical resistive MHD code with ionisation, coupled to the electrodes & circuit response. Then a two-fluid code to handle the filamentation and pinch. Followed by a PIC code for the plasmoid formation.

Once you get all that working for a simple deuterium plasma, only then can you move on to add the p+B11 ion species to the fluid stages, and including the strong magnetic field effect and radiative cooling to the PIC code.

I agree simulation of plasmas in general is not very advanced. Based on the current state-of-the-art codes for tokamaks & inertial fusion and how long they have taken to develop. I’d say it would take at least 10 years work by several reasonable size research groups for DPF models to get to even a comparable level of confidence.

I always like to bring up the comparison with steam engines – they were developed by people such as Watt in the 1770’s and improved by Trevithick in 1801, well before there was a good understanding of thermodynamics by the likes of Carnot in 1824. Certainly well before there was any way of modelling or theoretically calculating the efficiency of an engine design.

By experimenting we may crack fusion well before we know what’s going on at the microscopic scale. The trouble is that the scientific community has got used to building ever larger experiments on the scale of LHC, NIF, ITER, JWST etc. where the money men want hard evidence that projects on that scale will work before they part with the cash. We need more of the trial & error approach.

[Henning]

jamesr wrote: There are plenty of PIC codes that can handle the modest densities of a DPF plasmoid (ie around solid density) Inertial confinement simulations have to cope with 1000 times higher.

I meant here, simulations specifically for DPFs. Sure, the big guys have supercomputers where they can handle this, but the science of DPF is much too small for this currently.

[delt0r]

The term supercomputer is not a good one anymore. Sure some would say i don’t have access to one. Yet our cluster has 500 cores and all have at least 4gig ram per core. The high mem nodes have 16gig per core. This is more powerful that supercomputers 10 years ago. The way here in the EU clusters are shared with departments and just asking for time. Getting time on a cluster would not be difficult or expensive. I am sure the US has similar programs. I am even allowed to put on “fun” code on when its empty (often enough).

In fact one of the main reasons you are seeing a lot more simulation work with ITER class devices is that computers are finally catching up to the details that are needed (aka neoclassical transport and ELM etc). Pinches are however harder to simulate since there are very large gradients in just about every parameter. Using Vlasov type equations (guiding center stuff) is not easy or IIRC even possible.

[cccccttttt]

“Pinches are however harder to simulate since there are very large gradients in just about every parameter.”

Harder is a long way from impossible.

At one time digitally testing the theory of quantum chromodynamics was out of the question.

Eventually they built a dedicated computer for the job.

So too will there be a day when your services for FF designing are required.

[asymmetric_implosion]

Henning wrote:

There are plenty of PIC codes that can handle the modest densities of a DPF plasmoid (ie around solid density) Inertial confinement simulations have to cope with 1000 times higher.

I meant here, simulations specifically for DPFs. Sure, the big guys have supercomputers where they can handle this, but the science of DPF is much too small for this currently.

That is also not true. Sandia and Livermore are working on PIC modeling of the PF implosion phase. PF-1000 has been subjected to these models with limited success. A PF at Livermore is undergoing modeling using the LSP simulation. The results from Livermore are likely to appear at the American Physical Society Division of Plasma Physics meeting in a couple weeks.

The PF community in the US is larger than you believe. In fact, there are experiments from New Jersey to California at different scales and with very different goals. However, a superior understanding of the basic physics benefits all applications. Folks around the world see potential in PF technology for producing medical radioisotopes, material modification, radiation sources and weapons effect testing. Few people believe that the PF will ever produce net fusion energy for a number of reasons. Honestly, the main reason is a number of engineering problems like material lifetime. The PF has the same problem as ITER and NIF; you generate a hot plasma. That plasma expands and although it cools, it hits the solid materials. The PF electrodes are damaged a little on every shot. Thus, the electrodes have a finite lifetime. The use of copper like that is used on FF-1 is less than ideal. Published literature (Rout et al IEEE Transactions on Plasma Science vol 23. No 6 1995) showed that Tungsten based materials offer higher fusion yields than copper with far longer lifetimes. This is not to say that experiments shouldn’t be conducted. Materials are the limitation of all fusion devices. Materials are also the least funded fusion area in the US. As more and more experiments show promise of breakeven or gain, the more serious engineering challenges will be addressed.

[zapkitty]

Tungsten is also less than ideal as an electrode material because of the x-ray flux. While FF theory holds that brem will be limited enough to allow net power, the x-rays will still be quite strong. (thus the “onion”)

The most-discussed alternative is beryllium.

[cccccttttt]

“….The PF electrodes are damaged a little on every shot.”

You raise an interesting point.

How many shots will the electrodes perform before they degrade?

Certainly an important engineering question after a prototype is working.

It’s pure blue sky but one can dream of replacing the electrode with an

ultra stable EM field that would behave as an electrode.

The virtual electrode!

ct

[zapkitty]

Last I heard estimated FF electrode lifetime was about a month, revised downwards from an earlier estimate of about 3 months.

Given electrode recycling, wear isn’t expected to be a big impactor on cost or performance unless the electrode lifetime gets a lot shorter is than currently thought. An apt comparison would be the wear and maintenance costs of a gas turbine generator and associated gear… and that’s a comparison that the turbine just can’t win.

[Author]

Posts