superconducting magrid design

[jon117]

Hi guys,

I've been lurking for a while now and finally registered on the forums, so I thought it was time to share some of my ideas... I think Famulus had it right when he said we need to build a superconducting magrid. Go big or go home, as it were. So here's my design. This is a complete work in progress. I'm looking for your input guys. I'm trying to keep it so that it can be machined from solid blocks of stainless steel on a 3-5 axis CNC machine.

This is a 1m diameter coil, dodecahedral magrid design.

It should be able to produce around 5T @ 1kA, or 10T @2kA
64 coils of 2mm diameter NbTi wire in an 8x8 grid. Obviously this means a liquid helium cooling system.

These are held in place inside a square 1mm thick aluminium pipe (h=16mm w=16mm) (red)

The pipe is held in position (to counter any lorentz forces) by 5mm thick supports (yellow)

Surrounding that is the LHe pipe (60mm diameter)

This is bolted to supports on the inside of the outer pipe. (135mm OD with a radius of 500mm)
I'm not sure if this should be filled with LN2, or if it should just be thermal insulation. thoughts?

The upper outer shell has inset holes to fit the bolts into, which screw directly into the bottom half of the shell. a copper seal at this point should keep everything pretty air/liquid tight...

Nowhe inset holes I'm not sure about, I know I should be avoiding sharp corners, but I didn't really see any other way to do it. I kept the side facing the reaction smooth to try and prevent arcing. Anyone guess if that'd be enough? Or would I have to plug the holes?

I have the feed-thrus in my head. I can explain if anyone's interested but I didn't model them out since it's late...

Anyway, let me know what you think guys!

(also, I only modelled 1/5th of the pipe obviously because it's the same shape all the way around and I can just duplicate it around 5 times)

[jon117]

max upload is 3...

[mvanwink5]

Thank you, very exciting!

[hanelyp]

You want a vacuum gap for thermal insulation between the liquid helium and liquid nitrogen layers. So:

Superconducting wires, probably multiple turns, in a liquid helium vat.
Vacuum gap.
Liquid nitrogen vat.
Vacuum gap.
Cool Water cooling.
Vacuum gap.
Hot water, molten salt, or liquid metal cooling.

[ladajo]

It is also useful to consider that the cooling mechanisms can all be used to recover energy from the system.

[krenshala]

It is also useful to consider that the cooling mechanisms can all be used to recover energy from the system.

If you have to have it anyway, might as well make the best use of it you can.

[jon117]

You want a vacuum gap for thermal insulation between the liquid helium and liquid nitrogen layers. So:

Superconducting wires, probably multiple turns, in a liquid helium vat.
Vacuum gap.
Liquid nitrogen vat.
Vacuum gap.
Cool Water cooling.
Vacuum gap.
Hot water, molten salt, or liquid metal cooling.

thanks for this.. wow, that's going to make the diameter of the coils REALLY large. and also add a bunch more complexity to the feed-thru mechanisms... I thought I was pushing it at 13cm for a 1m diameter coil..!

So the vacuum gaps are *absolutely* necessary for a setup like this? Most of the stuff I read involving LHe setups said you could get away with a two stage LN2 -> LHe with a little bit of insulation (ie. some aluminium foil), but then I guess they weren't dealing with such a large heat load..

Is the last stage really necessary for a research setup? Obviously I know it would be for a proper reactor, but I was thinking of this more sort of in terms of someone with access to a 5-axis CNC and some metal working connections who could knock it out in their free time, hence trying to keep it as simple as possible with little to no welding. My mind started spinning when I imagined trying to cycle molten lithium out of the inside of a vacuum chamber!

Adding two extra stages and vacuum gaps wouldn't be too difficult design-wise, but I worry at that point you'd have 30cm cross diameter coils for a 1m coil diameter!! And from my understanding, that's waay too much metal in there. Not to mention you're getting into pretty ridiculous complexity for fabrication at that point too..

so i think we need a trade-off of complexity for size and ease of fabrication. would it be possible to do:

superconductor
LHe cooling
LN2 cooling
vacuum gap
water cooling?

Thanks for the help guys, please keep it coming!

[hanelyp]

A vacuum gap with polished reflective surfaces on both sides is about the best insulation you can get. The quality of insulation is independent of gap width, so you make it as small as practical given manufacturing and structural considerations. The multiple layers of coolant minimizes how much heat the inner layers have to carry off. Ideally all but the outermost coolant layer will pick up little heat steady state.

[ladajo]

Choice of material that interfaces with the vacuum gap is important as well. Especially in this construct where you would introduce significant neutron flux for D-T testing purposes.

[mvanwink5]

Have you given thought about feedthrough design?

[jon117]

yeah... the vacuum gaps threw off my initial thinking so I've gone back to the drawing board. I started modeling it out last night but didn't finish. It's muuuuch more complex than the initial design :S

basically, I need to be able to carry LHe, LN2 and the vacuum through flexible teflon tubes. Is this possible? if not, a steel design is doable.. but INFINITELY more complex to fabricate.

so far my design has it so that it looks something like this (sideways cross-section of the teflon tubes):

_______________________
vacuum 1
_______________________
LN2 2
_______________________
LHe 3
_______________________
LN2 2
_______________________
vacuum 1
_______________________


hope that's understandable... anyone know how viable it is to do this?

[mvanwink5]

These threads might be helpful in case you had not seen them:

for previous coil design effort:
viewtopic.php?f=4&t=1174

cooling requirements (magnetic field breakpoint drop for pB11 might be of interest):
viewtopic.php?f=4&t=2090

Purchase MRI coils plan (current price for used coils might be important):
viewtopic.php?f=4&t=1258

Just some reading to see some previous thoughts....

Thanks for your effort so far.

[jon117]

Thanks for the links, no I wasn't aware of them.
Feed-thrus I'm thinking will look like this:

Where the bolts are, the part that sticks out and holds the bolt goes all the way around the circumference of the shape.

Where the feed-thrus are, that part does NOT intersect the bolts, and does not go the full circumference of the piece. Those parts stick out (think jigsaw pieces) and friction fit into the inside of the next torus. Difficult to explain but hopefully you get the idea.

Please excuse the awful quality and hand drawn-ness. It's infinitely quicker to sketch these ideas out than to model them...

Thanks

[mvanwink5]

Jon117,
And the Sc?
I was thinking as an alternative, coming out the bottom with a nested pipe, through drilled holes and welded out as the next outer layer is added. The tops would have no through holes. The Sc could then be in the center of the nested pipe. (if that is understandable). Just a thought.

PS The other issue is if the SC needs to be electrically isolated from the outer coil metal case due to the high voltage of the coils.

PSS The water supply could be in a separate inlet and outlet pipe.

[jon117]

yes, i understand what you're saying, however this is insanely problematic from a fabrication standpoint. i initially thought about the method you described, but abandoned it. the way i see it is this:

you have the torii halves for the magnet all nested inside of each other, right? ok awesome.

then you drill holes and weld the feed-thrus for each progressive layer.

now, you can do one of two things (because remember, this is all going to have to connect seamlessly to the OTHER magrid(s))

1) you weld on feed-thrus of half the length required between torii.

2) you weld on feed-thrus of full lengths required between the torii.

1) is impossible. short of using some sort of x-ray welding. or possibly some sort of endoscope welding. you can weld the outside pipe, but then you can't get access to the pipes inside. because you've just welded up the only access point! you can't bolt the parts together because that would make the feed-thrus incredibly bulky, not to mention, again, the problem of not actually being able to access the bolt holes.

2) is.. incredibly complicated, and my thoughts tell me again impossible, because of exactly the same reason.

yes, if we were building just one magnet that way would work perfectly, but the nature of the geometry means short of using some extremely exotic fabrication techniques, like 3D printing the entire F*ing coil(!), the one i proposed is the simplest and easiest way to fabricate.

your other points:

the SC is labelled NbTi in my diagram and i only drew half of it. it is supported by standoffs inside the tube (see my initial designs - the yellow things).

my initial thought was to drive the magnet at low V at between 1-2kA. however my knowledge of superconductors is pretty limited. could you elaborate further please? what would cause the need for the coil to be driven at high V?

you're absolutely right, the water pipe could be separate. is there an advantage to this? I didn't really see a reason to separate it, so I left it to be partial insulation for the lower temp feed-thrus.