Hey Stefan, now, upon reflection i would have to agree. Those four pipes at the corners would 100% be the electron guns. The "fifth" pipe which i hadnt even noticed would have been the gas injection. Cheers ! It seems like there are two geometries where one could inject electons/ions either through the centre of the toroid and in the corners. I have fixated on just one in my Magnetron Thread
Shielding the electron guns using the "shadow" of the magrid is an excellent idea i too have considered. I guess the question is if the B field is strong enough to "curl" the electrons to where the B fields converge without screwing things up. The valencia paper and some of the early 90's articles in fusion technology go quiet deep in explaining the relationship between injection angle/well depth. The basic premise is that the injected electron beams should radially converge symmetrically (eg 12,3,6,9 oclock) at equal energies. If one starts injecting electrons at odd angles etc this genererates angular momentum in the well which decreases the well depth.
These alpha particles are starting to be nasty little buggers the more we think about it. Out gassing/contamination will be a problem. I am starting to think the answer might lie in diamond coating and low profile design of critical components.
But thats life hey. I saw a televised interview with one of the chief ITER guys, when asked about the neutron bombardment he said that simply the reactor walls would have to be replaced every couple of months. *laughs* what a dick.
Virtual Polywell
Yeah. I like that. The escaping charged particles (alphas) leave their electrons behind.
Initially the electrons can be injected with 2 MeV energy (assuming alpha decelerating grids) from the walls and you just accept the initial losses this causes. Once every thing starts to work you shut off the electron guns and it just works.
Initially the electrons can be injected with 2 MeV energy (assuming alpha decelerating grids) from the walls and you just accept the initial losses this causes. Once every thing starts to work you shut off the electron guns and it just works.
I figured a simple brute force calculation would be the easiest thing to do. So the first thing I did was figure out how much storage space a simple calc would take. For my parameters, it was a mere 256 TERAbytes. Since I've only got a few GB, it ain't gonna happen on my home machine!!
So I headed in the general direction of pencil and paper and started looking at the formulas. Again for crude and simple assumptions. It is interesting that I found for an assumption of a microwaved ionization core where the electron density is an exponentially decreasing function from the center out and the velocity distribution is Maxwellian I get a stable electron density distribution with time (expansion damps rather than explodes exponentially) when the core is hot is the outside is cold. I can also be stable if the electric field times the temperature (velocity distribution) is larger than the radial decrease in density.
In "reality terms", the electric field strength has to be larger than the expansion pressure. That the math agrees with common sense is a good sign, especially for a simple model
I think the next step will be to add in the electric and magnetic fields from the electron fluid. This will separate the polywell from the IEC. The present crude and simple model shows that the magnetic field does not change the electron fluid at all, it just drops out of the df/dt formula with the thermal distribution assumption. This is actually a good thing, because it says that a given electron distribution won't change by the external magnetic field. The assumption is a thermal distribution though, so it will be interesting to see how the math changes with the addition of an interacting field from the fluid itself.
I've got a ways to go before the "virtual polywell" model is useful, but so far it's been really interesting. And having a really powerful computer wouldn't hurt either, so it seems like a lot of fun all around!!
So I headed in the general direction of pencil and paper and started looking at the formulas. Again for crude and simple assumptions. It is interesting that I found for an assumption of a microwaved ionization core where the electron density is an exponentially decreasing function from the center out and the velocity distribution is Maxwellian I get a stable electron density distribution with time (expansion damps rather than explodes exponentially) when the core is hot is the outside is cold. I can also be stable if the electric field times the temperature (velocity distribution) is larger than the radial decrease in density.
In "reality terms", the electric field strength has to be larger than the expansion pressure. That the math agrees with common sense is a good sign, especially for a simple model
I think the next step will be to add in the electric and magnetic fields from the electron fluid. This will separate the polywell from the IEC. The present crude and simple model shows that the magnetic field does not change the electron fluid at all, it just drops out of the df/dt formula with the thermal distribution assumption. This is actually a good thing, because it says that a given electron distribution won't change by the external magnetic field. The assumption is a thermal distribution though, so it will be interesting to see how the math changes with the addition of an interacting field from the fluid itself.
I've got a ways to go before the "virtual polywell" model is useful, but so far it's been really interesting. And having a really powerful computer wouldn't hurt either, so it seems like a lot of fun all around!!
Perhaps look at a distributed computing approach, such as http://boinc.berkeley.edu
Yes, but scanning 256 terrabytes will take a long time even on a distributed system. Better to think a little harder and reduce the size of the problem. Brute force is nice if you can use it, but sometimes a little finesse can go a long way to getting the answer quicker. And sometimes just as accurately.
I suspect boinc will be a useful tool no matter what. Marketing the problem to the world is another chore. I think solving the world's energy problems is a worth while problem to work on, but solving the right problem and convincing people it is the right problem are two different tasks. I know I suck at marketing!
I suspect boinc will be a useful tool no matter what. Marketing the problem to the world is another chore. I think solving the world's energy problems is a worth while problem to work on, but solving the right problem and convincing people it is the right problem are two different tasks. I know I suck at marketing!
Yes, I think you could easily partition the problem for BOINC. Short range interactions can be done on each PC. Each time step consists of two parts, field equations and particle motions. Once you know where all the particles are, you can compute the long range fields, then figure out how all the particles interact with the fields, etc. Since each time step is a huge problem in and of itself, it takes a lot of computing to finish. I'm not sure how BOINC stores data, but setting up lots of terrabytes is not that hard these days. 30TB is $850, so it's really possible to create the central store.
But first, let's see if it is needed!
Fun to think about though - that's for sure!
But first, let's see if it is needed!
Fun to think about though - that's for sure!
I haven't gotten to POPS yet. One step at a time!
But I realized my cost estimate is way off - too late at night and not close enough reading. 30TB is closer to $18k. Still, it's accessable and possible to think about 256TB as a disk farm of servers with distributed world wide computing chomping on it and saving to DVD's for later display.
Bussard's nuclear rockets here we come!
But I realized my cost estimate is way off - too late at night and not close enough reading. 30TB is closer to $18k. Still, it's accessable and possible to think about 256TB as a disk farm of servers with distributed world wide computing chomping on it and saving to DVD's for later display.
Bussard's nuclear rockets here we come!
Typical, too late at night and too much fun. We'll have to wait a couple of years and then I'll be correct!
Computer simulation in those conditions with fidelity would be more challenging than building an experimental reactor. Besides simply increasing the B field is not going to compress the core like you think.DavidWillard wrote:How about a simulation of pulsed higher states of energy to the coils? There the exponential question.
Beta =1 is where magnetic pressure equals electron pressure. If you dont maintain relative symmetry between electron and magnetic pressure the well will blow out.
Purity is Power
I think the electric field is more important that the magnetic field. In toroids confinement uses the magnetic field parallel to the containment vessle and talking about beta=1 makes sense. For the polywell the magnetic field is mostly radial and perpendicular to the containment vessle.
The way I think about it, the magnetic field is what holds the electrons away from the surfaces, either coils or containment vessle. You want that field to be large enough that the electrons circulate as a current. They spend most of their time in the core and the rest of the time going up through the coils and back down through the cusps. With lower density far away from the core and high density in the core, the ions see the core "as the place to be".
The electron density can blow itself out of the core - the magnetic field helps it flow back again. If the circulating currents in the plasma equal the currents in the coils, then you can't hold them. In that sense beta=1 is a bad state to be in!
So far, I'm assuming the currents in the coils are much larger than the currents in the vessle. The math gets way too complicated otherwise!!
The way I think about it, the magnetic field is what holds the electrons away from the surfaces, either coils or containment vessle. You want that field to be large enough that the electrons circulate as a current. They spend most of their time in the core and the rest of the time going up through the coils and back down through the cusps. With lower density far away from the core and high density in the core, the ions see the core "as the place to be".
The electron density can blow itself out of the core - the magnetic field helps it flow back again. If the circulating currents in the plasma equal the currents in the coils, then you can't hold them. In that sense beta=1 is a bad state to be in!
So far, I'm assuming the currents in the coils are much larger than the currents in the vessle. The math gets way too complicated otherwise!!
This is a really interesting subject to me. I know some C++, but not much else. What language are you using? Fortran? (Does anyone even use that anymore anyway?)
Also, how are you thinking about incorporating the electric fields: as interactions between a set of particles that you keep track of, or by assuming some distribution (say spherical space charge with density ~1/R for example)?
Also, how are you thinking about incorporating the electric fields: as interactions between a set of particles that you keep track of, or by assuming some distribution (say spherical space charge with density ~1/R for example)?