thread for segments files and parameters for simulation runs

[happyjack27]

for my time scale multiplier constant i have:

((RC2/E_TO_C)*0.002*0.00000000025)

RC2 = 1/ (speed of light)^2.
E_TO_C = the elementary charge (charge of a proton) in coloumbs.

multiply that by the "time step" parameter in the top left and you can see the how many seconds each frame represents. though i've been varying it. but that gives you a ballpark.

the sim runs at about 15 frames per second, so multiply that and you get sim seconds per real second.

i'll settle on one for the 15cm sims after i run some tests, which may take a while.

[KitemanSA]

A picky point, but in this model the inner magrid did not cancel the field in the center. It had it's own field towards the center. bbets

Are you simply saying that the superposition was imperfect and resulted in some small residual field or that the image field so overwhelmed the real that there was a significant field inside the image magnet? I thought it was the former. Indeed, I was under the impression that the cancelation was nye on perfect. No?

[D Tibbets]

KitemanSA., I'm saying (I think) that the superposition of the internal magnet represents an outward force that excludes/ pushes out the original magrid B field, and that this creates the B-field surface topology or geometry that fits the expected topology of the presumed outward Plasma pressure mechanism. It confirms the claimed morphology of the Wiffleball, but it does not necessarily confirm that the Plasma would also do this. That reacquires experimentation with measurement and/ or dependence on understood plasma physics, possibly both neutral and non neutral flavors.

Dan Tibbets

[D Tibbets]

for my time scale multiplier constant i have:

((RC2/E_TO_C)*0.002*0.00000000025)

RC2 = 1/ (speed of light)^2.
E_TO_C = the elementary charge (charge of a proton) in coloumbs.

multiply that by the "time step" parameter in the top left and you can see the how many seconds each frame represents. though i've been varying it. but that gives you a ballpark.

the sim runs at about 15 frames per second, so multiply that and you get sim seconds per real second.

i'll settle on one for the 15cm sims after i run some tests, which may take a while.

Lets see, ...
RC2 is very small (1/ 30,000,000 cm/s or ~ 3 * 10^-8 cm/s)
E_to_C is very very small ( 1/ ~6*10^-19 Coulombs/ proton)
So RC2 / E_to_C=~ 1 *10^11 protons cm / Coulomb sec
Multiply by your constants gives ~ 1*10-3 protons cm / coulomb sec.
Now how do you cancel out the protons, cm, and coulomb units so that you end up with seconds only (or are those units implied in your constants)?

My head hurts. How about a final ballpark number instead of a formula?

Dan Tibbets

[erblo]

I think you forgot to take the inverse square of c - could be important

My guess is this (using the standard SI units):

RC2=1/(3 * 10^8 [m/s])^2 ~ 1.1 * 10^-17 [s^2/m^2]
1/E_TO_C=1/(1.6 * 10^-19 [C=As]) ~ 6.24 * 10^18 [1/As]

This gives:

((RC2/E_TO_C)*0.002*0.00000000025) ~ 1.1 * 10^-17 * 6.24 * 10^18 * 5 *10^-13 [s/Am^2] ~ 3.4 * 10^-11 [s/Am^2]

Using the numerical part of this gives a time step of 34 ps, which is then multiplied by the time step multiplier set by the slider.

(A 10keV electron has a speed of ~ 6 * 10^7 m/s (I think) and would therefore move something like 2 mm during one time step.)

Does this sound like something you would do happyjack27?

[happyjack27]

k, sorry about that.
i'm using much higher precision for my constants.
if i did my calculations right just now, RC2/E_TO_C is about 69.446160587277...
and the remaining part is of course
0.0000000000005
multiply together and you get:
3.4723080293...e-11

so essentially what erblo said.

to get the seconds per second you still have to multiply by the frame rate (~15 fps) . which makes:
5.208462...e-10 sim seconds per real second

i'm going to change it and run a few trials. ultimately it will be in orders of 10 on a second scale. e.g. 10^-12 seconds per second.

also, thanks, erblo, for the translation to mm! that time step is too large for the 15cm radius version. i actually end up multiplying it be <1.

[happyjack27]

i'm running a sim now at 10 picoseconds per second (~1 picosecond per frame) 15cm radius, ~0.82 tesla, 10E-6 net ion charge, 1% excess electrons. i plan on recording a few hours, then speeding it up w/video editing software ("time lapse"), 10x and 100x. then doing the same for 1 picosecond per second, and comparing.

then if they're close, i'll post the time-lapsed videos, otherwise i'll take it down another order of magnitude, and when i get a close pair i'll post it.

[happyjack27]

...

Also, shrinking the size of the system may help. Things would evolve faster in a 1 mm wide Polywell than in a 1 M Polywell. Of course these manipulations may also, greatly decrease any accuracy you currently have.

Dan Tibbets

shrinking it down seems the best way to go, given that my dominant limiting factor is the particle count.

not sure i'd be comfortable going all the way down to a mm, though. my spatial resolution is the resolution of a single precision floating point number, w/1.0E0.0 = 1 meter. so the minimum value is on the order of 10^-38. well, okay, maybe my discomfort is unjustified.

but my main questions are thus thusly:
*how do i determine what the proper mag field strength, at, say 1cm radius? it scales linearly w/the radius, right? so 0.8/15 tesla?
*how do i determine the ion population count?
*how do i determine the electron population count? presumably, ion count, plus a certain well depth?

EDIT: aye, turns out a proton is about 10^-15 meters in size (radius?). looks like my discomfort is completely unjustified! at this scale i can fit as many protons as you could fit in a cubic football field inside each individual proton!

[KitemanSA]

KitemanSA., I'm saying (I think) that the superposition of the internal magnet represents an outward force that excludes/ pushes out the original magrid B field, and that this creates the B-field surface topology or geometry that fits the expected topology of the presumed outward Plasma pressure mechanism.

Ok, I think we are basically saying the same thing. When the fields superpose inside the image magnet, the fields cancel (or approximately so) but outside the add in such a way that they simulate the effect of the electron pressure. And that is the point of this exercise, simulation. No?

[happyjack27]

i would think they would cancel in both directions, but only at a radius between them does it reach net zero field. inside that there's residual field from the inner magrid, outside, residual from the outer magrid.

on that note, if the theory is correct, would not another virtual magrid form inside the virtual magrid to cancel _its_ field? and so on... resulting in a sort of fractal?

EDIT: erg, nevermind. just read up on the method of images. my thinking was off. inside they would be additive. and the method of images wouldn't apply recursively because the plasma is not "superconductive" relative to itself.

[KitemanSA]

EDIT: erg, nevermind. just read up on the method of images. my thinking was off. inside they would be additive.

Yup, inside they are equal but opposite thus add to zero (ish). Between, the image radial gets weaker and the real radial gets stronger with increasing radial position so they cancel but not totally. The tangential all head in the same direction so are additive CONstructively....

[icarus]

It is a spherical inversion of the field about the spherical boundary (the wiffle-ball radius).

Whatever the physical field looks like exterior to the physical Magrid will be reflected to be the field interior to the image MaGrid, albeit scaled by the spherical inversion factor. (An infinite radius reflects back to the origin.)

Similarly, the physical field between the physical Magrid and the spherical boundary will be reflected (i.e. inverted) to be the same as the image field between the image Magrid and the spherical boundary.

At the spherical boundary, the radial field component contributions from the physical and image MaGrids are equal and opposite and thus form the spherical boundary with the zero normal flux condition.

[rcain]

sorry to come so late to the thread. (been busy).

Firstly, HappyJack and RJP - Brilliant work! congrats so far - you are up there with the mighty Indreck and beyond. IMHO.

Coupleof thoughts:

1) I wonder if we can work with Famulus to set some real operating points and calibrate your sim?

Is it easy to set up some virtual instruments/probes?

2) Wiffleball formation - I dont see where we gain by further considering 'analageous' (virtual mirror) models. As HJ has already pointed out, surely at this point we want to see the WB emerging 'naturally' from the basic physics implicit in the sim. As already suggested, shrinking it should reveal what we want to see. What factors are missing?

3) love the phase space views, but realy need some 'orientation'/directions to ensure i'm interpreting correctly what i'm seeing. Any chance of something (datum) on-screen.

at some point would be very interested in further investigating one of your earliest vids - showing response to ('inadvertant') step function.

4) is there a way to get slightly higher res vids. i 'just' cant make out what the sliders are/are set to.

5) Am I correct in assuming your intention is to get the sim as 'complete' as possible a representation of the Plolywell rig (and regimes): ie. Charged Magrid, Injectors, (Divertors?), Collisions, Nuclear forces, cross sections, start-up, steady-state, etc?

Thought of hiring your time to Doc Nebel and team?

respect

[KitemanSA]

HAPPYJACK27:
Simple question. Do your simulated moving charged particles create a simulated magnetic field? Yes or no?

[happyjack27]

Is it easy to set up some virtual instruments/probes?

depending. i can set up some point charges. to do any totals i'd need to do a "parallel reduction" and/or "atomic add"s. so that's not so easy. other phase space views are fairly trivial at this point, though. now that i got the framework set up, i just throw in the calculations.

As already suggested, shrinking it should reveal what we want to see. What factors are missing?

i'm finishing up my time step experiments first. but what i'll need is:
1. magrid radius (circumscribed sphere)
2. magrid current (amp turns) (i can convert from teslas at coil center to amp turns if need be)
3. ion population count
4. electron population count

as to ion population count, that's the major limiting factor. i'm not comfortable with going past a total charge of about 10E-6. i might push it if need be though. since a single ion has charge about 10E-19, that would be about 10E13 ions. and i only have about 10E4 points in the system to represent the ions, so that's an enormous representation ratio as is.

my spatial resolution is that of single precision floating point, so we can go pretty small if need be. and the current is also single-precision floating point so it's practically unlimited, too.

1.5mm?
if someone can tell me what my population counts and coil current should be, i'll do it.

love the phase space views, but realy need some 'orientation'/directions to ensure i'm interpreting correctly what i'm seeing. Any chance of something (datum) on-screen.

i could add some segments to the magrid with 0 current 0 charge 0 width.

4) is there a way to get slightly higher res vids. i 'just' cant make out what the sliders are/are set to.

for the past few videos and every video from now on, i'm just uploading the raw .wmv youtube then converts it to various resolutions: 240,360,480, & 720p. on the lower right of the video you'll see an place where you can choose what version to watch.

5) Am I correct in assuming your intention is to get the sim as 'complete' as possible a representation of the Plolywell rig: ie. Charged Magrid, Injectors, (Divertors?), Collisions, Nuclear forces, cross sections, etc?

right now i'm focused on accuracy and data acquisition. i already have a cross-section view in phase space. (the very last one). but there are other things on my to do list, involving completeness:

1. revisit my static e-field formulas and hopefully fix them (charged magrid)
2. now that i know a proton's radius is well within my precision range, it's trivial to just subtract it from the distance in my e and m field calculations, thus treating it as a charged sphere instead of a point charge.

i have injectors ("guns"), i just don't have them calibrated to the right velocity. so instead i'm simulating injection by ionization. though really there are no "neutrals" (as far as i know), the "ionization" is just randomly introducing new ions and electrons. if there are to be neutrals i don't want to add them ad-hoc, i want them to form naturally via already simulated em-forces. and thus also ionize naturally the same.

nuclear forces - i.e. weak force and strong/residual strong force. the scale is too small for my simulation range. they're not really relevant at this scale anyways. residual strong force is simulated via the nuclear cross-section calculation (and with it also quantum tunneling, etc.)

collisions - i've seen this word thrown around in this context, but i'm not really sure how it comes into play. the particles just bounce off via coloumb forces, right? they never actually rally "collide" (touch), and if they did, well that would be a fusion event, right? so as far as i kow in simulating inter-particle em-forces i'm simulating "collisions" in this sense.

Thought of hiring your time to Doc Nebel and team?

would love to. if they're interested, i'm available.