superconducting magrid design
Re: superconducting magrid design
I think it's a decent idea, but I'm just worried about insulation for the LHe. how about we have the LN2 go all the way around, but make the last vacuum chamber include the back half of the toroid?
so you'd have the plasma facing side with water in it, the back side would be a vacuum, then you'd have full pipes with: LN2 (separated from the water by a vac gap), vac, LHe?
Does that make sense? It cuts the number of nested pipes down to three. I think it's an acceptable trade-off and I can probably figure out a way of separating the LHe and LN2 feed-thrus to make it even simpler.
I'll be checking in here less since I just got put on crazy overtime at work for the next few weeks, but I'll still be checking in and doing some modeling whenever I can.
Oh, and about the venting thing, I'm not so sure that's a great idea. 1x10-9 is pretty annoying to get to even on small chamber sizes, and venting those enclosed spaces (likely using small holes) will just create outgassing. also you're then exposing the coils to a much higher vacuum than they would otherwise have been at, which means their surfaces need to be forged steel, polished smooth, etc. etc... it just raises the complexity bar even more.
so you'd have the plasma facing side with water in it, the back side would be a vacuum, then you'd have full pipes with: LN2 (separated from the water by a vac gap), vac, LHe?
Does that make sense? It cuts the number of nested pipes down to three. I think it's an acceptable trade-off and I can probably figure out a way of separating the LHe and LN2 feed-thrus to make it even simpler.
I'll be checking in here less since I just got put on crazy overtime at work for the next few weeks, but I'll still be checking in and doing some modeling whenever I can.
Oh, and about the venting thing, I'm not so sure that's a great idea. 1x10-9 is pretty annoying to get to even on small chamber sizes, and venting those enclosed spaces (likely using small holes) will just create outgassing. also you're then exposing the coils to a much higher vacuum than they would otherwise have been at, which means their surfaces need to be forged steel, polished smooth, etc. etc... it just raises the complexity bar even more.
Re: superconducting magrid design
Sounds reasonable. Hope they pay you for the overtime.
Counting the days to commercial fusion. It is not that long now.
-
- Posts: 1439
- Joined: Wed Jul 14, 2010 5:27 pm
Re: superconducting magrid design
there are, as with any engineering problem, conflicting goals in the design of a superconducting polywell. i find it often helps to enumerate.
* inner volume should be large
* grid should be thin (bigger means more electron loss and lower mag field strength at casing - since mag field dies off at r^2)
* grid should be mechanically stable
* high mag field - as high as possible
* very high vaccum
* grid should be able to handle lots of power throughput, as well as voltage (power non-throughput)
...
some questions come to mind:
* do you really need 3 cooling layers - he, n, h20?
**how cold does the superconductor really need to be?
**how hot is the chamber really going to get?
*the above questions depend on the material - what material for the superconductor? ytrium-based?
**...consider the conductivity of the material at different temperature - and the trade-off between mag field strength and size ( given that more cooling = stronger mag field, but bigger grid).
*would it benefit or hurt to lace the superconductor with sapphire to encourage magnetic "pinning"?
* inner volume should be large
* grid should be thin (bigger means more electron loss and lower mag field strength at casing - since mag field dies off at r^2)
* grid should be mechanically stable
* high mag field - as high as possible
* very high vaccum
* grid should be able to handle lots of power throughput, as well as voltage (power non-throughput)
...
some questions come to mind:
* do you really need 3 cooling layers - he, n, h20?
**how cold does the superconductor really need to be?
**how hot is the chamber really going to get?
*the above questions depend on the material - what material for the superconductor? ytrium-based?
**...consider the conductivity of the material at different temperature - and the trade-off between mag field strength and size ( given that more cooling = stronger mag field, but bigger grid).
*would it benefit or hurt to lace the superconductor with sapphire to encourage magnetic "pinning"?
Re: superconducting magrid design
I don't have Sc pricing, but that is likely the most expensive component of construction. For standard NdTi wire, the matrix includes copper for quench protection. The lower the temperature, the higher the current carrying capacity of the Sc. We should have a comfortable margin for grid protection from both electrons and alphas based on earlier heat loading board discussions (assuming 5T toroids or higher). I am fairly sure the LN is needed to practically use LHe as it seems to be standard practice. Mechanical strength is a big deal to support the large magnets with high B field. This is what Billy said a while back:
First, let me apologize for my long absence. I had some paying work come in and I had to concentrate on it. Second, for those who missed it, I am making jackets and supports of the SC core from 5083-O aluminum, now, not stainless. This grade of aluminum gains strength as the temperature drops and it contracts at the same rate as the SC coil. That is two very large advantages all by themselves, but even more important is the enormous reduction in weight. I propose to use that material for the LHe core, the first vacuum jacket, the LN2 jacket and the 2nd vacuum jacket.
I propose to make the cool water jacket out of Inconel 690. It is very high strength and has zero cobalt in it. It machines and welds about as easily as the austenitic stainless steels. It is not nearly so prone to work hardening as the other nickel alloys like the Hastelloys. It is one of the preferred materials used in light water fission plants and holds up quite well to neutron bombardment.
I think a Inconel 690-copper-Inconel 690 laminate for the Hot Water jacket, Kieth calls it the "first wall", is quite doable. I think I have come up with a way to make the longitudnal seam welds on such a jacket that would hold as well as if we were seam welding solid Inconel 690. However, such a joint would need to be fabricated and tested for us to be certain.
Kiteman is right about my transferring the mechanical loads through the jackets to the vacuum vessel walls. I am still working on the six big solenoids locked in a sumo contest and do not see any other way to deal with the loads with that configuration.
Chief Simon, I wonder if you would be good enough to think about this stuff:
http://en.wikipedia.org/wiki/Yttrium_ba ... pper_oxide
If we could make Bitter Plates with this material we could eliminate the need for LHe. I don't know if that is in the realm of possibility, but it would make life a lot simpler if it is.
I have managed to squeeze out a little time for this project here and there and have finished with the trickier pieces. I hope to have more pictures for everyone to look at soon.
Thank you all for your patience,
Billy Catringer
Counting the days to commercial fusion. It is not that long now.
Re: superconducting magrid design
Thought I should mention that we need to be talking some different dimensions for the Sc:
B = u x j / 2R, u = 4pi x 10**-7 T m/A, (B in tesla, j in amps, R in meters)
1 meter in diameter, R=0.5
B=10, then j = 10 x 0.5 x 2 / (4 x pi x 10**-7) = 8 million amps (or amp turns)
10**5 amps / sq cm means need 80 sq cm
or 9 cm x 9 cm Sc for 10 T toroid of MgB2, 1 meter in diameter.
reference posts:
magnetic field calculator link supplied by msimon:
http://hyperphysics.phy-astr.gsu.edu/hb ... urloo.html
MgB2 conductor capacity
viewtopic.php?f=4&t=1215
B = u x j / 2R, u = 4pi x 10**-7 T m/A, (B in tesla, j in amps, R in meters)
1 meter in diameter, R=0.5
B=10, then j = 10 x 0.5 x 2 / (4 x pi x 10**-7) = 8 million amps (or amp turns)
10**5 amps / sq cm means need 80 sq cm
or 9 cm x 9 cm Sc for 10 T toroid of MgB2, 1 meter in diameter.
reference posts:
magnetic field calculator link supplied by msimon:
http://hyperphysics.phy-astr.gsu.edu/hb ... urloo.html
MgB2 conductor capacity
viewtopic.php?f=4&t=1215
KitemanSA wrote:I recently ran across an article that implied 10E5A/cm^2 for MgB2 at LHe temperatures. That would make the core more like 10cm radius. The abstract of the paper from Japan reads:MSimon wrote:We can start out with a 10 Tesla coil 2 meters in dia. 16 million ampere turns. Figure a Jc of 10,000 A/sq cm That is 1,600 sq cm. That makes a radius of 29 cm just for the superconductors.
{{Up caret added due to lack of superscript in this software.}}The paper reports the first successful fabrication of MgB2 superconducting tape using a flexible metallic substrate as well as its strong pinning force, which was verified by direct measurement of transport critical current density. The tape was prepared by depositing MgB2 film on a Hastelloy tape buffered with an YSZ layer. The Jc of the tape exceeds 10^5A/cm2 at 4.2K and 10T, which is considered as a common benchmark for magnet application. The Jc dependence on magnetic field remains surprisingly very small up to 10T, suggesting that the tape has much better magnetic field characteristic than conventional Nb-Ti wires in liquid helium.
Last edited by mvanwink5 on Thu Mar 21, 2013 5:32 am, edited 1 time in total.
Counting the days to commercial fusion. It is not that long now.
Re: superconducting magrid design
hey guys, just popping my head in with a quick note.
i priced out the NbTi wire needed for the magnet a few weeks ago.
- 64 turns (8 x of 2mm - NbTi wire with copper
- for all 12 magnets (something like ~650m I forget the exact number)
- It was capable of running at 2kA
- has an operating temperature of ~9K
total mag field (theoretically) was ~10T. Not sure if I'm calculating this correctly though.
total came to: $4000.
so not as expensive as you might think OK I have to get back to work... oh, and yes, they're paying me for the OT!
i priced out the NbTi wire needed for the magnet a few weeks ago.
- 64 turns (8 x of 2mm - NbTi wire with copper
- for all 12 magnets (something like ~650m I forget the exact number)
- It was capable of running at 2kA
- has an operating temperature of ~9K
total mag field (theoretically) was ~10T. Not sure if I'm calculating this correctly though.
total came to: $4000.
so not as expensive as you might think OK I have to get back to work... oh, and yes, they're paying me for the OT!
Re: superconducting magrid design
jon117,
Bad news I am afraid. See previous post. Sc based on $4k for 128kamp-turns will be $250k because 8 million amp-turns are needed for 10T coils (for each coil).
Bad news I am afraid. See previous post. Sc based on $4k for 128kamp-turns will be $250k because 8 million amp-turns are needed for 10T coils (for each coil).
Counting the days to commercial fusion. It is not that long now.
Re: superconducting magrid design
Remember that the field the SC has to endure is much higher than the field at the center of the coil.
For a thick linear conductor the magnetic field is 0 in the middle and increases linearly to the full radius of the conductor. The field outside is the same as for an infinitely thin wire carrying all of the current. A first order approximation of a coil (toriod) is then to look at the off-axis field of the current loop, for example http://www.netdenizen.com/emagnet/offax ... ulator.htm, and take this as the field up to the surface of the coil (the current carrying SC, not the casing). This can then be used as the maximum field the SC has to withstand. This approximation becomes worse for thicker toroids.
For the example of a main radius of 0.5 m and 8 MA the field in the centre is ~10 T but it's ~20 T at the surface of the coil if the minor radius (radius of the SC bundle) is 10 cm. For 5 cm it will be closer to 40 T.
For a thick linear conductor the magnetic field is 0 in the middle and increases linearly to the full radius of the conductor. The field outside is the same as for an infinitely thin wire carrying all of the current. A first order approximation of a coil (toriod) is then to look at the off-axis field of the current loop, for example http://www.netdenizen.com/emagnet/offax ... ulator.htm, and take this as the field up to the surface of the coil (the current carrying SC, not the casing). This can then be used as the maximum field the SC has to withstand. This approximation becomes worse for thicker toroids.
For the example of a main radius of 0.5 m and 8 MA the field in the centre is ~10 T but it's ~20 T at the surface of the coil if the minor radius (radius of the SC bundle) is 10 cm. For 5 cm it will be closer to 40 T.
-
- Posts: 1439
- Joined: Wed Jul 14, 2010 5:27 pm
Re: superconducting magrid design
...but those are calculators for a conductor, not a super-conductor.
with near-perfect conductivity, i would think the magnetic field inside a superconductor would be near-zero, and then change rather abruptly at the surface. i'm just presuming here though, not an expert here.
with near-perfect conductivity, i would think the magnetic field inside a superconductor would be near-zero, and then change rather abruptly at the surface. i'm just presuming here though, not an expert here.
Re: superconducting magrid design
But they can only exclude up to a certain strength at which point they stop being superconductive.
Re: superconducting magrid design
Assuming a single turn, circular cross section, uniform current density: the magnetic field is zero at the center of the wire, increasing linearly with radius from the center to a maximum at the edge of the wire. If the cooling jackets weren't an issue, making the coil minor radius a tight fit inside the casing would be favored to minimize the field the superconductor need handle.
To first approximation the same applies to a compact multi-turn coil as would be used for the magrid.
To first approximation the same applies to a compact multi-turn coil as would be used for the magrid.
The daylight is uncomfortably bright for eyes so long in the dark.
Re: superconducting magrid design
erblo, point taken, thanks (nice calculator). So the critical field for the Sc must be able to handle 40 T if a 10 cm diameter Sc wire bundle is used. I suspect unless jon117 wins the lottery or his firm pays well, $250k will slow the 10 T coil construction plans a bit.erblo wrote:Remember that the field the SC has to endure is much higher than the field at the center of the coil.
For a thick linear conductor the magnetic field is 0 in the middle and increases linearly to the full radius of the conductor. The field outside is the same as for an infinitely thin wire carrying all of the current. A first order approximation of a coil (toriod) is then to look at the off-axis field of the current loop, for example http://www.netdenizen.com/emagnet/offax ... ulator.htm, and take this as the field up to the surface of the coil (the current carrying SC, not the casing). This can then be used as the maximum field the SC has to withstand. This approximation becomes worse for thicker toroids.
For the example of a main radius of 0.5 m and 8 MA the field in the centre is ~10 T but it's ~20 T at the surface of the coil if the minor radius (radius of the SC bundle) is 10 cm. For 5 cm it will be closer to 40 T.
Counting the days to commercial fusion. It is not that long now.