Convergent Scientific videos are up.

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

Moderators: tonybarry, MSimon

Post Reply
mattman
Posts: 459
Joined: Tue May 27, 2008 11:14 pm

Convergent Scientific videos are up.

Post by mattman »

Hello,

Sorry for the multiple posts, but these videos have lots of material.


Building a Polywell - 68 min Investor Pitch:
http://sproutvideo.com/videos/7c9bd8bc1a11e4c7f4

Numerical Modeling of a Polywell - 29 minutes
http://sproutvideo.com/videos/e89bd8bd1314edca60

Commercial Applications - 22 minutes
http://sproutvideo.com/videos/189bd8bd131be6c290

mattman
Posts: 459
Joined: Tue May 27, 2008 11:14 pm

Re: Convergent Scientific videos are up.

Post by mattman »

Hello,
I watched this presentation. I made some notes on it. I did some research on what I did not know and wrote it up. I am looking for corrections, insight and feedback. I have also left some questions I would like to answer.

Starting on slide 4: “How plasma is confined”

Image

I see two principals here:

1. A better vacuum helps performance. Air in the chamber will cool off the plasma.

2. Hotter plasma helps performance. He mentions three methods for heating:

a. Radiofrequency heating – this is like microwaving the plasma. Once it is hotter than 16 eV, the deuterium breakings into (+) and (-). The Lockheed effort may use this.

b. Induction heating – I assume he means when a magnetic field is switched back and forth, causing gas to heat up and ionize. Though it is hard to see the difference between this and RF heating.

c. Hot particle injection – this is what it sounds like.

A couple questions:

1. What is the electron energy inside the device? Is it different from the ion temperatures? Any data on this?

2. Why does he not mention heating ions using electric fields? Isn’t this the fundamental principal of fusors, polywells and all IEC devices?

3. I am unclear on the difference between inductive heating and RF heating. They sound very similar; maybe someone can explain this to me?



Slide 5: “Confinement is hampered by anomalous transport”

Image

The goal in magnetic confinement is to hold in the plasma. However, in all cases, the plasma is never contained nearly as long as predicted (10x or 100x short time frames). The reason given is Anomalous transport. As far as I can tell this is a blanket term for all the physics caused by instabilities and it gets worse as machines get bigger. There are lots of papers on anomalous transport, especially in the tokamak world.

Slide 6: “4 ways to stop anomalous transport”

Image

Anomalous transport (AKA plasma instabilities) is suppressed by:

1. Velocity shear – this is too vague. What does he mean? Does he mean differing rates of flow within plasma? Does he mean differing directions of flow within plasma? Does he mean higher velocities of flow within plasma? Is a plasma “swirling” around the outside, going to suppress instabilities?

Image

2. Steep plasma gradients.

Image

3. Small surface to area volume ratios.

Image

4. Sharp field gradients.

Image

Slide 12: “Problems with fusors”

Image

Four problems with fusors:

1.Energy loss as material hits the cage and is conducted away.
2.In big fusors, cage heats up and burns off.
3.Electrons are lost when they bounce off one another or the electric field and hit the wall.
4.Ions thermalize in the core and/or mantle. I take Mr. Baker to mean the same core, plateau, mantle and edge model that Tim Thorson worked out. This is from page 27 of his thesis. If so, I have tried to mark out where thermalization is taking place.

Image

Slide 13: “Variables which control the polywell”

Image

I think this slide deserves more emphasis, for a number of reasons:
1. Many skeptics argue that the plasma will not have this much fine structure, AKA specific “edge” and “core” properties. IDK - I do not think there is hard data on this. The closest data I have seen was Joe Khachans October paper, which basically measures electron trapping at low beta.

Image

This data shows some edge and core behavior, but it is not conclusive. To measure this, I think we will need Thompson laser scattering of a polywell. You shine a laser into the plasma, read the reflections and get maps of the densities. Wisconsin recently (October) added this to their HOMER fusor.

2. I am not sure this is a complete variable list. We need a good list of the input parameters and dimensionless numbers. Here are some I like:

Image

Picking dimensionless numbers is an art form. Once you pick good ones, you can make sense of your data, guide experiments better and map out modes of operations with fewer simulations. If I had cash, I would do this for the polywell. Here is a description of their list.

Image

He mentions a couple points. First, that edge density determines overall cloud stability. Second that edge electrons motion “resists” the applied field. Skeptics would pounce on this. Though, there is a physical mechanism and evidence from magnetic mirrors of this; we do not have strong published data showing this in polywells. This forms the basis of the “whiffle ball” confinement concept. He cannot make that statement unless he has the data to back it up.

Slide 16: “Problems we have found”

Image

This is the meat of the presentation. Here is what he mentions:

1. Ion injection is a problem. Dr. Alex Klein, who worked for Bussard listed ion injection as a major problem. You need to control: (1) location, (2) energy, (3) scattering and (4) vacuum. Joel may have circumvented this issue with his “differentially pumped ion source” which is a mechanical solution that is mentioned later on but is not explained. They even provide a picture of this device. Does anyone have a better picture, where you can make out these labels?

Image

2. They mention two acoustic instabilities: Weibel and Diocotron instabilities. I am weak on instabilities; if someone can explain these to me I would appreciate it. Here is my understanding of Weibel.

Image

This instability also occurs in a counter streaming case. Some questions:

a. Does this occur in (+) beams and (-) beams? Does it occur in “packets” of moving neutral plasmas?
b. What is the difference between the counter streaming case and the uniform beam case?
c. What are the factors that lead to it or stop it? Beam speed? Beam density? Beam diameter? Background environment? I see that it can be created with a beam is hit with an off-axis electromagnetic perturbations.
d. Where does this apply in the polywell? In the lecture I watched a beam was sent into neutral plasma and a counter stream formed against it. Though the polywell plasma is very (-) I think this would happen in a polywell.


The Diocotron instability is created when two sheets of plasma move past one another. Here are some cases:

Image

Are all these cases true? Can this be mitigated? Where would this be an issue inside the polywell?

3. Mr. Baker says that the magnetic fields must be “conformal” to the metal surfaces, both at startup and at the end. The field cannot cross a metal surface after the plasma pushes the B field around. It sounds like they are working hard on this.

Image

4. Loss of convergence inside the device. They give a few reasons:

Image

a. Lensing or defocusing of plasma at the edge. What is the physical mechanism of “lensing”?
b. Ion collisions cooling down the ion population.
c. Electric and magnetic heating of the ions at the edge.

mattman
Posts: 459
Joined: Tue May 27, 2008 11:14 pm

Re: Convergent Scientific videos are up.

Post by mattman »

5. Loss of the potential well by “well washout”. This appears to happen overtime as the ions are injected. He mentions Joel Rogers working on this. He gives a ball park number: that the well is 10% of the outside electric field voltage.

Image

He mentions screening effects and a cold ion population created by charge exchange as the reasons. A couple of questions:

Image

a. Why do cold ions do more screening that hot ions? I am guessing that hotter plasma contains more voids of space inside.

b. They observed that this is less of a problem in the cusps. Why? Are the electrons in the cusps hotter? Is this why, in WB6 they put the emitters at the corners of the device?

c. Does he have any numbers on this? How hot are the ions? How hot are the electrons? What is the theoretical or simulated
rate of energy exchange from the two clouds? Rider wrote one paper on this subject where he estimated the rate of energy exchange between hot ions and cold electrons.


6. Non- adiabatic region extending into the cusps. “…When you have this cold electron population formed by collisions and diocotron oscillation, this can cause the potential profile in the gap to be very flat, with a very thin skin depth. This allows the non-adiabatic region to extend into all the way through the cusps, causing a loss of confinement...” I am not sure what he means by this, some questions:

a. Where is “the gap”?

b. I would love some kind of graphic showing what they think is happening to the potential profile over time…

c. What are the non-adiabatic and adiabatic regions? Is he referring to the regions that Joe Khachan talked about in his 2011 paper?


Image


Slide 18 “solutions to our problems”

Image

They are starting from pure electron plasma. This is really not surprising. Differentially pumped ion sources are mentioned, but not explained. It is a method to keep the ions hot and from colliding with one another to stay hot. Fast response external bucking coils. These are extra magnetics behind the regular rings which allows them to control the electron and ion population which is outside the rings. This is scraping and wave heating. A couple of questions:

a. The most distinctive part of CSI design, is these second magnetics. I do not like them. I am interested in doing more researching on specifically what a “bucking” coil is, when they are used, how effective they are and what is the cost (in conduction losses) for having them.

They close with pictures and questions. There is lots of interesting facts in the questions.

Image

Image

prestonbarrows
Posts: 78
Joined: Sat Aug 03, 2013 4:41 pm

Re: Convergent Scientific videos are up.

Post by prestonbarrows »

Just watched the presentation. I'll try to answer some of your questions.

-To first order, the electron energy at the center (in eV) is just the voltage applied to the grid. In reality, this is broadened into an energy spectrum by collisions. Neglecting collisions, a particle's energy simply trades between potential and kinetic with the total remaining constant.

-Ions are ~1000 times more massive than electrons. Since electrons have such low mass, Lorentz forces accelerate them around much more easily. Almost all of the power you pump into a plasma through electric or magnetic fields gets coupled to the electrons (and then possibly transferred to other species by collisions or space charge effects a.k.a. the 'virtual cathode'). It is not really possible to directly heat ions alone using EM fields. Think of a bunch of ping-pong and bowling balls in the breeze.

-RF (radio frequency) heating is an umbrella term for any scheme that couples energy to a plasma by applying fields that oscillate in the RF frequency band. Inductive heating (often) uses RF frequencies. Basically you have some type of coil creating an oscillating field that sets up a secondary current in the plasma. It is exactly like a transformer with the secondary 'coil' formed by currents within the plasma. RF can also be used in other geometries such as two parallel flat plates where only the electric field is important and there is no significant inductive coupling.

-Tokamak and instabilities are a bit outside my experience but I believe velocity shear refers to a velocity gradient from the center of the torus towards the edge. i.e. the core circulates faster then the edges near the wall. The plasma generally circulates around the major diameter of the torus.

-I believe the picture is of an electron gun not an ion injector. Seems to say 'E-gun' on a few of the captions.
These systems are what I am most familiar with. From the junk quality image a few key things to point out: grading rings between the coil and wall help prevent arcdown at higher voltages; these are basically a resistor divider ladder which holds a fixed field gradient through space and smooth out the electric field. It appears they are showing some type of filament with an electrostatic lens which would accelerate/focus the electrons into a beam.

-Curved E-fields cause focusing effects on charged particle beams. There will be curved E-fields between the coils and chamber. I believe this is what they mean by 'lensing'.
http://en.wikipedia.org/wiki/Electrostatic_lens

-Electric field screening in a plasma is described by the Debye length. The electric field due to a given charged particle is effectively screened at locations more then a few Debye lengths away. This length scales as (temperature/density)^0.5. Basically, a colder and denser plasma means you need less distance to get the same number of screening particles because the charged particles are packed closer together.
http://en.wikipedia.org/wiki/Debye_leng ... n_a_plasma

-Regardless of screening, cold ions from charge exchange will 'fill up' the virtual cathode. They would push the center more towards neutrality which reduces the well depth without a significant gain to fusion rates.
http://en.wikipedia.org/wiki/Poisson's_ ... trostatics

-'adiabatic' refers to an adiabatic invariant. In terms of a plasma, these are three specific values which remain constant for a single charged particle orbiting in an external magnetic field with no collisions (since B-fields can do no work). They cause a given particle to follow a well-defined path; often trapping it on a closed surface. However, a magnetic field is required for these to be valid. At the center, where |B| goes to zero, the invariant is no longer defined and the particle scatters in a random direction. So adiabatic regions are those with a significant B-field near the coils with well-defined motions and non-adiabatic is the low B-field region at the center with chaotic motions.
http://en.wikipedia.org/wiki/Adiabatic_ ... ma_physics

-I work with deferentially pumped ion sources. It is a very well developed concept. Forms of such systems vary wildly but they are used in particle accelerators, industrial processes, and ITER among others.

-An ion injector looks very similar to what is pictured. Instead of the filament, you have a small plasma source and the polarity of the voltage on the lens is flipped to pull ions out. The plasma source requires a relatively high pressure so you would locate it further away on a tube with turbomolecular/cryo pumps between the source and polywell to block gas conductance into the grid region.

-It is definitely an interesting option. I think the hard part will be tuning the beam energy correctly to get the ions up and over the grid potential 'hill' into the virtual cathode without the beam blowing up due to space charge or simply plowing through to the opposite side. It would only really be feasible if you had a very deep virtual cathode well. Though this application is very different than the beamline systems I work on.

mattman
Posts: 459
Joined: Tue May 27, 2008 11:14 pm

Re: Convergent Scientific videos are up.

Post by mattman »

Ok.

At the end of the presentation they mention: "Model I". This is a device they really built, starting in 2010 and ending in the Spring of 2012. They trapped electrons with it, presumably measured by Langmuir probe. Here are the tests he mentions:

Voltage from cage to rings: |...| Measured voltage drop in center:
1,000 volts |...| 600 volts
1,500 volts |...| ?
10,000 volts |...| 900 volts

Model I has single turn copper magnets. They cool it with Fluorinert at high pressure and a second water-gycol coolant system. It is unclear if this has 6 or 8 faces. Its similar to PXL-1. They designed it based on a "parameter space" given by simulations. This means they have already done some operating space simulations. Mr. Blaire gives some more details:

Image

Does anyone have a photo of this "Model 1"?

===
He mentions that "Model 2" was included in Joel Rogers 2012 presentation... here are the pictures Joel used.

Image

prestonbarrows
Posts: 78
Joined: Sat Aug 03, 2013 4:41 pm

Re: Convergent Scientific videos are up.

Post by prestonbarrows »

These show their 'hybrid' mode with a magnetic grid around a standard IEC style grid. Its not clear if that is what they were referencing with those numbers. Looks like a different setup though judging by the '1.b' label.

Image
Image

mattman
Posts: 459
Joined: Tue May 27, 2008 11:14 pm

Re: Convergent Scientific videos are up.

Post by mattman »

Right, the team has built two devices of interest: Model 1 and Model 1B.

Model 1:

Model one is a diamond copper wire structure. Here is what it looks like:

Image

Image

I made a film, showing this field:

http://www.youtube.com/watch?v=yHGO2Fbs ... e=youtu.be

What I do not know is the size of everything. There are photos, but estimating the size from these has not shown to be accurate. Here is what I have been able to tell from CSI documents (and presentation).

Image

At that current the wire would overheat. They added a fluorinert coolant layer to accompany the wire into the cage. But, I am not sure if the coolant layer is on the inside or outside of the wire. Which would be better? You can apply Biot-Savart to do easy estimates of the field.

Image

Using these numbers and these equations, there should be an easy way to estimate Model 1. Model one places fields – pointing in opposite directions – next to one another. This is different from WB6; where the fields were parallel. They built this with the idea that the mirror effect was independent of direction. These opposing fields should change plasma behavior; but I am not sure how yet. You can see this below:

Image

Model 1B:

Here they have added a negative inner cage. This is a fusor, with a magnetic field outside; a fusor/polywell hybrid. It is important to note how novel this is - it is very new.

Image

This is what the fields look like inside (in an off-axis plane):

Image

I would imagine the electrons would cluster around the corners of the magnetic cage, away from the inner electric grid.

mattman
Posts: 459
Joined: Tue May 27, 2008 11:14 pm

Re: Convergent Scientific videos are up.

Post by mattman »

Hello,

Ok. I am working on a write up now. Here is a section of the rough draft. Let me know if you have feedback.

======

Redesign:

People have long wondered about redesigning the containing magnetic field. They argue that the magnetic mirror effect is independent of direction. The mirror will reflect particles when they move into denser fields [10]. This happens only under certain conditions; but it will happen regardless of if the field points one way, or the other [XXXX]. Redesigning has been a source for much speculation. At last check, over 850 posts online debated different designs [15, 9]. Endolith, an electrical engineer, has argued for the new design of an 8-sided spherical cage [16]. Alternative designs have even been patented - going back to the 1980’s [18, 19, 20].

Keys to “Good” Fields:

In reality, many magnetic field designs will fail to hold in particles. They look good on paper, but will fail in reality. Failure hinges on the structure of magnetic fields inside. We do not understand everything about magnetic containment. However, there are some principals that can guide our efforts:

1. Uniform Strength: Electrons in the center should see uniform field strength in all directions. This is like water held in a pool. The walls of the pool must be the same height. This was the driving force behind WB6– the design, forced the corners and the sides to have equal field strength [XXX].

2. Continuous: All the fields must not run into a metal surface [XXX]. This drives up conduction losses and was the reason for HEPS failure.

3. Not Intersecting: Fields which stream particles past one another - lead to counter-streaming, two stream and diocotron instabilities [XXX].

4. Low Curvature: Bent fields, fling plasma outward [XXXX]. This effect worsens with tighter curvature.

5. Symmetry: This has been true throughout most of magnetic confinement. In polywells, this means that the same fields cannot butt up against each other. An inward pointed field cannot be next to another inward pointed field. This is shown best with two simple geometries: six rings in a cube and an eight sided diamond. Notice that no red surfaces meet. Each design can be expanded by adding more rings or loops.

Image

Bussard patented the diamond configuration first [21, 22]. The 1989 patent features a prominent image of an octahedron. CSI has built this. They say they have run current through it and, trapped electrons within it [17].

Model 1:


This design is known as model 1. It is very odd looking, when compared with the WB6 rings. First, it is made only by one wire; it is a “single turn” magnet. The change has tradeoffs. It hurts the device, because it requires hundreds of times more current to get the same fields. But, it helps the device, because it reduces the amount of extra surfaces (struts, connectors, feed lines) in the center. The wire bends are notable. It is likely - that the most interesting and problematic physics occurs there.

Image

CSI has indicated that several experiments were attempted. These included several materials, such as copper and tungsten [6]. Some designs included cooling tubes inside [14] and it is possible that geometries beyond the octahedron were attempted. The company will not disclose the exact dimensions of model 1. However, from the parameters provided in their presentations [6] and on their website [23], we can discern a range of sizes.

Basic Math:

CSI list the average confinement radius for model one as 16 centimeters [23]. It is unclear what they mean by this. This may be the mean distance from cage to device center. If this is true, model one is twenty-six centimeters per side. Alternatively, they give the plasma volume as 1.4E-3 cubic meters [23]. If this is the total volume inside the cage; then model one is fourteen centimeters per side. Finally, they list the maximum field strength as 1,000 Gauss [23]. The strongest fields should be in the center; but to reach this with their maximum current of 1,500 amps – the device would need to be seven centimeters a side. These calculations were found by geometric and magnetic analysis, the equations are shown below.

Post Reply