You are here

IEC Simulation

This project will probably be ongoing for a while, since I'm just poking around here and there when I get spare time. But it's pretty interesting already, so I'll just post some screen shots here as I make progress.

In case you missed it, here's the blog where I introduced this: Inertial Electrostatic Confinement (IEC) Fusion

Core Designs


This was an early mock up of the polywell fusor design that Dr. Bussard came up with. The red and blue faces show the polarity of the magnetic fields, which you can clearly see by the yellow arrows indicating the current direction, notice how each face has a wind order that corresponds to the color of the face, and there aren't contradicting areas, like you would get if two like faces were touching, as you can kind of see by the small area in the first image. The last 2 images were just me experimenting with the design in maya.

Magnetic Field Lines


Here is one of my first simulation screenshots. I made it so the yellow lines are current lines which I can place however I want, and the rest are depicting the magnetic field lines. This was interesting to note how the field lines tend to bunch at the center of each face, and how they form a star in the middle, which you can't tell from that view angle. These field lines were all made by firing a streamer that follows the magnetic field lines which start from equally spaced grid points distributed throughout the cube.

Magnetic Field Lines 2


Next I wanted to see the actual shape of the field lines more closely, so I made a cursor I could move around that was a single starting point I could use to examine the field. Also I had been detecting when field lines loop in a funny way and I didn't like it much so I tried to revise it here. I was having a LOT of trouble though because I was expecting the field lines to loop and I was getting large discrepancies. I thought maybe it had to do with how I was stepping through the field lines using an adaptive RK solver. That wasn't the case, it was actually very accurate. Eventually I let the ends keep going and going and to my amazement they did in fact loop, after spiraling around a toroidal volume deformed by the field configuration.

Toroid Field Lines


This next image I made using a simpler wire arrangement to demonstrate the simplicity of the idea that field lines don't make short loops, they often have to spiral around a toroidal surface a few times before looping. If I just had the one coil (ring) each field line would pretty much meet right back up where it started without traversing any sort of volume. But it becomes clear with a wire running through the middle causing a twist to the whole thing, that they have to travel around several times before they can come back to where they started.

Toroid Field Lines 2


I noticed the lines even after going around a bit don't meet up precisely where they left, and over time fill more and more space on the surface of this toroid, and I thought it looked neat so here's another shot with more lines.

Electron Confinement


Finally on to some more serious stuff. Here I've started simulating electrons. At this point they are moving due to the external magnetic fields. They tend to curl along in a helical path around the field lines which are like guiding centers. The force is F = qVxB/m, where V is the velocity and B is the magnetic field, q is the charge, m is mass, and x is cross product. This is only one component of the lorentz force, but it was quite interesting to see how it would work to keep electrons confined in the middle. That was really my early goal in fact, simply to see how well the electrons stay stuck in the middle, and this was a great success. The integration was kind of tricky to figure out too due to the spiraling nature I made a custom integrator that breaks it down into the linear part and the centripetal part, and kept using adaptive RK stepping even for a specific render dt. This means my framerates can kind of explode. I didn't care much though I was more concerned with accuracy than framerate. With only a single threaded app, running on 1 core, with no simd instructions, or gpgpu calls, I was only about to simulate about 10 particles at a time. The dt was about 1/60,000,000th of a second, which was needed just to be slow enough to watch the electrons move around. And this was before any particle-particle interactions too!

Iron + Electron Confinement


Here I mixed in some ions with the electrons to see how the difference in charge and mass would look in parallel with the electrons. It was interesting to watch how particles escaping the system found ways to turn themselves around as well, and this is before any coulomb forces were added. It's entirely due to the magnetic field interaction!