thread for segments files and parameters for simulation runs

Discuss how polywell fusion works; share theoretical questions and answers.

Moderators: tonybarry, MSimon

happyjack27
Posts: 1439
Joined: Wed Jul 14, 2010 5:27 pm

Post by happyjack27 »

i got the eguns working!

http://www.youtube.com/watch?v=qe_hWCdr_bQ

very tricky to get beta=1. i've done it once, but the eguns just werent accurate enough so they just kept getting deflected, while the inner core stayed stable as ever. beautiful. addictive now.

icarus
Posts: 819
Joined: Mon Jul 07, 2008 12:48 am

Post by icarus »

happyjack:
very tricky to get beta=1.
How did you determine that? i.e. where/how are you outputting (measuring) magnetic field pressure and kinetic pressure to determine beta?

Pretty pictures, any chance to put in transluscent Magrid or simply a cubical box wire-frame for the faces that the coils lie upon? Might give the motions some more context.

Looks like fun ... got any heating problems on that gpu card yet?

happyjack27
Posts: 1439
Joined: Wed Jul 14, 2010 5:27 pm

Post by happyjack27 »

re beta=1, from the simulations i've figured out what it means exactly. the efield from the charge on the coil makes the electrons want to accelerate towards the coils. the bfield make them (and the ions, for that matter) want to go _away_ from the coils. as far as i know (i don't know for sure the scaling laws of magnetic reflection), the two accelerations both scale w/ 1/r^2 (mag field strength does, at least).

so where the mag field acceleration is stronger than the efield acceleration, the electrons will be concentrated in a sort of star mode -- the center of a bunch of convex fields pushing together. when the e-field acceleration is stronger than the b-field, you get sort of the inverse. right in between you get a good equipotential sphere. by definition, then, (of equipotential) anything that reaches zero KE anywhere inside it stays inside it. forever.
Last edited by happyjack27 on Sun Nov 21, 2010 8:00 pm, edited 1 time in total.

happyjack27
Posts: 1439
Joined: Wed Jul 14, 2010 5:27 pm

Post by happyjack27 »

a stable core:

http://www.youtube.com/watch?v=ajNef9v0sUs

i accidently jacked the magnetic field up suddenly.
problem is it's a two way street -- or roadblock, rather. electrons can't get out OR IN. it seems the thing to do, ideally, is create your electrons inside the well. through ionization or something like that.

happyjack27
Posts: 1439
Joined: Wed Jul 14, 2010 5:27 pm

Post by happyjack27 »

re beta=1, from the simulations i've figured out what it means exactly. the efield from the charge on the coil makes the electrons want to accelerate towards the coils. the bfield make them (and the ions, for that matter) want to go _away_ from the coils. as far as i know (i don't know for sure the scaling laws of magnetic reflection), the two accelerations both scale w/ 1/r^2 (mag field strength does, at least).

so where the mag field acceleration is stronger than the efield acceleration, the electrons will be concentrated in a sort of star mode -- the center of a bunch of convex fields pushing together. when the e-field acceleration is stronger than the b-field, you get sort of the inverse. right in between you get a good equipotential sphere.

icarus
Posts: 819
Joined: Mon Jul 07, 2008 12:48 am

Post by icarus »

happyjack
so where the mag field acceleration is stronger than the efield acceleration, the electrons will be concentrated in a sort of star mode -- the center of a bunch of convex fields pushing together. when the e-field acceleration is stronger than the b-field, you get sort of the inverse. right in between you get a good equipotential sphere.
kind of seat of the pants and rough around the edges but in essence correct, at least how I think of it anyway.

Since the electric potential is related directly to the kinetic pressure, then the balance between the drive electric potential and the magnetic field of the coils is related to beta ... beta is defined as the ratio between magnetic pressure and kinetic pressure, measured at some point in the field ... the shape of beta=1 surface is what we are looking for .... if mag. field dominates it will be much more star like and vice-versa if the electric drive potential is stronger then the beta=1 surface will balloon out into a more spherical shape

time to send in the ions?

happyjack27
Posts: 1439
Joined: Wed Jul 14, 2010 5:27 pm

Post by happyjack27 »

icarus wrote:
time to send in the ions?
looks like we're about there. i want to maybe try getting the eguns the right distance to achieve 0 KE in the center. then i'll add an i-gun featue and send those in with a little extra umph.

problem with that though is the grid is positively charged, and while that's great for getting the electrons in there, the ions i'll probably have to start in closer. if i start at the other side of the voltage hill and just use higher KE's, they'll have enough KE to climb the voltage hill on the other side of the grid, so they'll just go right out the other side. it seems the way to do it in practice, if you can't get the ion guns in close enough (and maybe even if you can), is by ionizing neutrals.

on a simulation level i should note the ions are MUCH slower. and i have to set the time scale according to the faster particles to get good enough temporal resolution for the fast and small gyrations. that's why i switched to all electrons. i figured since the ions weren't doing much at that time scale anyways, might as well get a clearer picture of the electron behavior. i think that picture's pretty well formed now.

and the bigger the e and m fields the faster the electrons go and the more i have to pullback the time scale. so now we're going to have very slow ions. i think i'm going to have to record a video, letting the simulation computer run overnight or something, then speed up the video some how.

alternatively, i can find the conditions for a good electron core, then just simulate that with a point charge w/a charge-to-mass ratio of zero so that it doesn't move. then it's just simulating ions in the presence of a point charge and a static e-m field. that and it's like 2000x faster.

icarus
Posts: 819
Joined: Mon Jul 07, 2008 12:48 am

Post by icarus »

When you introduce the ions the charge distribution is obviously going to be affected so the e-gun position for zero KE maybe changed also ...?

Having the ions in there may (will?) affect how well the electrons remain in a confined configuration so simulating electrons as a point charge would not be very informative unless you already know that is how they will behave when the ions are there (and later fusing? also) ....

You need to keep an excess of negative charge (electrons) to create the central well.

Compute time just escalated huh?

happyjack27
Posts: 1439
Joined: Wed Jul 14, 2010 5:27 pm

Post by happyjack27 »

icarus wrote:When you introduce the ions the charge distribution is obviously going to be affected so the e-gun position for zero KE maybe changed also ...?
yeah, i figured that. partly why i want to get the eguns positioned first. i can set up some code to constantly measure the central voltage and adjust the egun position accordingly.
Having the ions in there may (will?) affect how well the electrons remain in a confined configuration so simulating electrons as a point charge would not be very informative unless you already know that is how they will behave when the ions are there (and later fusing? also) ....
yeah. good points. be kind of a fail to cop out now after setting up a fully analytic all-pairs simulation, only to ultimately do what could be done w/paper and pen.
You need to keep an excess of negative charge (electrons) to create the central well.

Compute time just escalated huh?
yeah. it's been progressively escalating with higher field strengths. not looking forward to the higher times w/ions in it. i want to get a good hold of the electron dynamics so i'm not wasting so much time on the electron+ion runs. but i'm starting to get itchy to put them back in. :)

icarus
Posts: 819
Joined: Mon Jul 07, 2008 12:48 am

Post by icarus »

happyjack:
i want to get a good hold of the electron dynamics
I think you are doing great on this front so far (given all the assumptions and limitations that have gone into it) ... if nothing else we are getting some kind of picture how the electrons behave under different parameters.

I particularly liked seeing the ExB drift rotation of the electrons about the cusp axes both inside and outside the Magrid ... important feature to capture in my opinion.

Also there was that intriguing radial pulsation of the electrons in the ultimate core along the cusp axes .... I wonder if that is physical, what sets the frequency of it, is it always there, is it density dependent ..??? So many questions.

happyjack27
Posts: 1439
Joined: Wed Jul 14, 2010 5:27 pm

Post by happyjack27 »

icarus wrote:happyjack:
Also there was that intriguing radial pulsation of the electrons in the ultimate core along the cusp axes .... I wonder if that is physical, what sets the frequency of it, is it always there, is it density dependent ..??? So many questions.
the pulsations happen when i change the current or esp. voltage on the grid. or when the eguns introduce a burst of new electrons. it stand to reason that their frequency is directly proportional to the voltage gradient they're seeing. in time they all die out through dissipation. and i imagine on human time scales this would be practically instantaneous.

but i suppose if oscillated the voltage / current in tune with the frequency (which presumably you could calculate from the voltage gradient), you'd get a sort of resonance due to the viscosity created by the particles' magnetic field. so you'd get sort of a "POPS" effect.

happyjack27
Posts: 1439
Joined: Wed Jul 14, 2010 5:27 pm

Post by happyjack27 »

my plan:

i added the i-gun code in. and figured out where i'm going to place my iguns and eguns. for the mixed particle simulation run. here's the plan:

i will simulate a wb-6 style polywell (32-gon coils), w/ 3m coil radius, 3.3m distance from coil midplane to center. (giving a sizable coil separation). my eguns will be just inside of the coil midplanes, probably around 3.25m i'll experiment with that first to get them positioned right.

then i will be sending in the ions. my iguns, or gun, rather, will be a hollow sphere in the center, w/ radius just under half the magrid radius. so about 1.60m. this will be a rough simulation of ionization of neutrals through cyclotron resonance. (though i don't imagine it would be nearly as precise. or spherical, for that matter).

i choose this point because according to:

http://www.mare.ee/indrek/ephi/pef2/

it's right at the peek positive electric potential, so it won't affect the convergence radius of new electrons.

i suppose as a simulation of ionization, it should be starting off equal parts electrons here. that would take some code modifications. this will have to do for now. and i'm not all that worried about the ions escaping anyways, so there won't be a lot of new ones being introduced, anyways. and the equal part of zero is zero.

the only problem remaining is that i only have like 14k particles, so my space charge is like zero. i can make each particle represent many particles. but we're talking like 10^(19) particles per coloumb. and i'm not all that comfortable pushing that parameter past a few orders of magnitude. i like it best at 1.0.

so to compensate i'm going to run the coils at very low charge.

i figure i have 14336 particles, and with a net neutral proton-boron mix, 6/8 of those are electrons, so 10752 electrons w/ ~1.6x10^-19 coluombs a piece = a total of ~1.72266x10^-15 coloumbs.

now the question remains how exactly do you go from this to this?

from the looks of it, it'd ideal for me to start the eguns off w/nonzero KE, and/or much further from center. and also it'd be better (and more faithful to reality) to introduce electrons slowly rather than all at once. both of these would require substantial code modifications. and the former would require some calculations, namely converting electric potential to KE and that in turn to initial electron velocities.

happyjack27
Posts: 1439
Joined: Wed Jul 14, 2010 5:27 pm

Post by happyjack27 »

crap i was looking at field magnitude rather than potential.

i think for now i'm just going to settle for a rough-and-ready model: i'll use an egun in the center at 0 KE.

in the meantime, can anyone tell me: given two sense points which tells me the electric charge at those point, in coloumbs -- giving those two values -- (and assuming a straight path between points) how do i calculate the initial velocity in m/s i need to send an electron off from point a so that it stops exactly on point b?

happyjack27
Posts: 1439
Joined: Wed Jul 14, 2010 5:27 pm

Post by happyjack27 »

parameters for upcoming simulation run (mixed particle!):

_grid_
configuration: cuboctahedron, circular coils (32-gon approximated)
coil radius: 3m
coil midplane distance from center: 3.3m
coil current: 200k amp-turns
coil charge: 1 coloumb??

_particles_
egun zero-energy point: 2cm
igun zero-energy point: 2m
electron count: 12544 (87.5%)
proton count: 896 (6.25%)
b11 count: 896 (6.25%)
excess electrons: 7168 (50% of total population)
net charge (in coloumbs): -1.1484401*10^-15

i'm adding code to automatically scale the time step inversely to the field strengths, so i should be able to vary field strengths arbitrarily without having to futz around w/ the time scale.

erblo
Posts: 29
Joined: Thu Mar 18, 2010 1:08 pm
Location: Sweden

Post by erblo »

Hi everyone. I've been lurking on the forum for quite some time, but this is my first post so please be patient if i say something stupid or make a mess of the formating :)
happyjack27 wrote:
in the meantime, can anyone tell me: given two sense points which tells me the electric charge at those point, in coloumbs -- giving those two values -- (and assuming a straight path between points) how do i calculate the initial velocity in m/s i need to send an electron off from point a so that it stops exactly on point b?


I think you need to rethink that question. The potential (~voltage) at a point/line charge is infinite. That's also the answer to your question of how to relate the charge on the magrid to the voltage - you can't. You can however ask what the charge should be to produce a potential U at some point close to the charge (or for example a short radial distance from a line charge).

Post Reply