Page 1 of 2

Bikeshedding superconductiong magrid design

Posted: Sat Jul 20, 2013 12:42 am
by Kregus
Things I don't know, and haven't been able to find out:

Series VS Parallel on the magnets: I figure Parallel has more math involved to make sure the plasma core is in the center of the magrid. It also requires more overall power, but it vastly simplifies the superconduction cooling design.

Someone, and I'll try to find the post again, was talking about the inefficiencies of sharp angles, such as bolt recesses, on the side facing the plasma core. Would filler for the recesses and ceramic dipping of the polywell/rings work to fix this? I mean it would add diameter and distance from the core, but it would smooth out, ideally, any sharp edges.

Has anyone even tested trying to cool copper coils, not to superconducting levels, but just chill the living shit out of them and measure it, to test out the various theories on best practice for layered cooling would even work?

I, admittedly, have no idea what I'm doing, these are just thoughts I had while driving to work. Play devils advocate, tell me why I am wrong.

If you have other unfounded untested ideas, please post them, I'll see if I, and ideally others, will see if they can come up with reasons to punch holes in them. Hopefully we'll find something useful.

Re: Bikeshedding superconductiong magrid design

Posted: Sat Jul 20, 2013 1:37 am
by KitemanSA
The sharp angles issue has to do with suppressing Paschen Arcing and is of concern for any material that is conductive and at a high voltage, i.e., the entire MaGrid.

I believe it is the opinion of most of the people on this forum that WB8 is cooled with liquid nitrogen. It is the only real way to explain the fact that the B field is 4X what scaling would account for. It also explains the fact that they have a rather large LN flask at their facility.

Not SURE on this second one, but "convincing circumstantial evidence".

Re: Bikeshedding superconductiong magrid design

Posted: Sat Jul 20, 2013 1:44 am
by hanelyp
Wiring the magnets in series vs. parallel should take the same magnet power. Series needs more voltage, parallel needs more current.

Coolant flow I'm thinking should be parallel, independent of whether magnet power is series or parallel. The coolant won't be following the wires around anyway in many realistic designs. The power for a single magnet may be 2 wires coming in on a single support, and numerous turns around the ring. Coolant I figure is better handled coming in on one support and flowing single pass to another support to exit.

Sharp edges anywhere on the magrid casing is asking for high voltage discharge. Properly shaped to follow magnetic field lines is good for this. The casing needs to be conductive so it can be charged.

Copper (or most metal) wire has increasing resistance with temperature. Copper coils cooled with liquid nitrogen has been discussed around here.

Re: Bikeshedding superconductiong magrid design

Posted: Sat Jul 20, 2013 2:07 am
by Kregus
I'm not arguing that we, meaning whoever tries this, should try to get copper down to superconducting temps, but to use LN or LH, if available, to see if the designs proposed on this forum for cooling are valid or just theory. At the very least, it defrays the cost of superconducting cable/ribbon/wire, which as I understand it, is quite es'pansive.

The coolant flow can only follow the magnet design though, if it's parallel, then the cooling has to follow that design, or risk being conductive itself at such high voltages. If series, then you'd have to cool even the connecting portions of the wire, unless you somehow connected them with non-superconducting cable/wire.

Setting the magnets in parallel also adds math, like I said, to getting the plasma in the center of the polywell. I don't know how it would be affected be being off by X, Y, or Z. I'm just saying eventually, for optimal efficiency, if it's required, someone will have to take into account all the resistances of n magnet rings.

Re: Bikeshedding superconductiong magrid design

Posted: Sat Jul 20, 2013 7:13 pm
by hanelyp
The magrid as a whole is at several kV relative to the electron injectors and the surrounding cage, but low sustained current. The voltage driving the magnets is far less, but much higher current. As I recall the WB-6 magnets were powered by a bank of batteries. Deionized water could be used as a coolant in direct contact between wires at the magnet drive voltage. With individual turns insulated a conductive coolant presents no shorting problem.

Each individual magnet in the magrid is going to have enormous of amp-turns of wire. Making the conductors a tube through which coolant flows is going to present massive resistance to the coolant, unless you go for enormous amps through just a few turns.

Re: Bikeshedding superconductiong magrid design

Posted: Sat Jul 20, 2013 8:57 pm
by Kregus
I don't see how you could use deionized water to cool the wires connecting the magnet rings without introducing some crazy engineering problems.

Unless you propose each polywell ring have an in and out from the vacuum chamber and you run the attaching wires out of and back into the chamber to get to the next ring, but even then it would be easier to just keep use the same coolant and basically build a layered piping system that goes in and out of the chamber to get to each ring and just run the same type of cable. Unless there's a problem when you get superconductors that long.

I imagine there are trade offs between amp-turns vs amps, cost being one of them.

Re: Bikeshedding superconductiong magrid design

Posted: Sun Jul 21, 2013 5:08 pm
by ladajo
The other issue with copper at high fields is its malleability. Lots of force on it.

I have mused a few times on using small diameter copper tubing with a conductive coolant running inside. Make each coil with multiple turns, tightly packed to help eliminate movement, and maybe even a dip and bake for the whole construct. Then put it in a conformal can or coating.
One of the issues with cooling is evening it out to minimize mechanical stress. The above method would help with that.

Re: Bikeshedding superconductiong magrid design

Posted: Mon Jul 22, 2013 4:50 pm
by DeltaV
How about a modified Bitter Plate design, with larger central hole and square envelope (when cutting across one side of the donut) to fit a toroidal casing?

Image

Might aid cooling and also be conformable with a thin-film type of superconductor.

Re: Bikeshedding superconductiong magrid design

Posted: Mon Jul 22, 2013 7:33 pm
by hanelyp
Much better illustration of the bitter magnet, and brief description of water cooling, at http://www.ru.nl/hfml/about-hfml/levita ... -solenoid/
The de-ionized water is in direct contact with copper, and if conductive would short out the magnet. Note that bitter magnets tend towards high current and low impedance.

Re: Bikeshedding superconductiong magrid design

Posted: Tue Jul 23, 2013 3:50 pm
by DeltaV
Yes, much nicer pictures.

Image

Image

Larger major-diameter/smaller minor-diameter toroids will start to lose the structural advantages.

Maybe use a nonconductive, nonmagnetic, coolant-tolerant 3-D space grid for added support in the unused volumes exterior to the square cross-section windings and interior to the circular cross-section toroidal casings.

Re: Bikeshedding superconductiong magrid design

Posted: Tue Jul 23, 2013 4:23 pm
by ladajo
So how do you bend a bitter construct into a toroid?

Re: Bikeshedding superconductiong magrid design

Posted: Tue Jul 23, 2013 4:28 pm
by DeltaV
You don't.

You cut out flat circular arcs and stack them helically.

You enclose the resulting square cross-section torus in a round cross-section torus.

Re: Bikeshedding superconductiong magrid design

Posted: Wed Jul 24, 2013 1:04 am
by KitemanSA
Why bother? The fields you can stuff into a torus are not strong enough to call for a bitter magnet.

Re: Bikeshedding superconductiong magrid design

Posted: Thu Jul 25, 2013 4:12 am
by DeltaV
Superconductors included?

Re: Bikeshedding superconductiong magrid design

Posted: Thu Jul 25, 2013 7:41 am
by KitemanSA
DeltaV wrote:Superconductors included?
Yes. MRI main magnets are about the size if a WB-D magnet. The only thing that keeps them from being stronger is that the material itself can't take a stronger field. AFAICT, strength is not an issue.