Is There an Optimal Size for Magrid Casings?

[D Tibbets]

The real world consideration is 1 MW/sq m and about 100 deg K delta T. Inlet of the cooling loop to outlet.

1.4 meters from a 100 MW machine center, each sq meter will receive about four times that figure (100MW/(4*pi*(1.4m)^2)

Still doable?

Just to clarify hopefully (at least for myself), my estimate of the Magrid heat exposure follows. I'll use a square form facter for simplicity. A cube 3 meters on a side, has 3m^2 per side * 6 sides= 54 m total area. With 100 MW each m^2 would receive 2MW of heat. The actual Magrid surface exposure is only a fraction of this, but the ratio should hold (I think).
But only the inside facing half of theMagrid is directly exposed to the heat, while the outer side is not. Would this effectively half the heat/ m^2 number? And, if a significant portion of the 'heat' is penitrating neutrons, with a significant portion of them passing entirely through the Magrids, will this reduce the heat/ energy that needs to be absorbed (assuming the superconductor could survive a significant neutron flux)?
Once pass the magrid, the volume of the vacuum vessel will be significantly larger, so heat load should be considerably less.


Dan Tibbets

[MSimon]

The heat was computed as 20% of reactor output as alphas. One face. Of course the heat load peaks at the crown. On the edges you will get some alpha deflection and there is a lot more material. I did not take this into account and assumed an even load. If the outermost material is copper or plated copper the "back side" should add 10% or 20% to the heat capacity.

Neutrons are an interesting case. Best for the superconductors if they are unshielded. By using a thermalizing fluid (water) for cooling the outer sections of the SC magnets you definitely want to absorb all the thermal neutrons. A thin layer of B10 will do that (1 to 3 mm).

I believe that with a system like that SC lifetime in a D-D reactor can be made adequate (1 year minimum) possibly even optimum (10 years or more).

Of course more detailed calculations will need to be done with a proper solver. And then experiments. The above is strictly back of the envelope.

[KitemanSA]

...
Neutrons are an interesting case. Best for the superconductors if they are unshielded. By using a thermalizing fluid (water) for cooling the outer sections of the SC magnets you definitely want to absorb all the thermal neutrons. A thin layer of B10 will do that (1 to 3 mm).

So, you want B10 for the shield, but B11 for the MgB SCs. Hmmm. Cool! Natural boron might just portion out about appropriately.

[Billy Catringer]

Prior to and during fusion there will also be more protons striking the magnet shells. Proton strikes will manifest as heat even though no single proton will be as hot as a 4He nucleus. I think the boron coating is a very good idea, but you are going to see some cracking and spalling from flexure.

There will still be a problem with the inner faces of the tori growing faster than the outward facing tori. Heat induced expansion on one side will tend to warp the entire structure outward from the reactor center. It might be possible to compensate for this by making the outward facing halves thicker than the inward facing halves, but you would likely want to do a finite stress analysis on the structure to see if that trick would work. I would be afraid that the inward facing sides will buckle. Then again, it might be possible to limit or eliminate the buckling by giving the inward facing shells a number of smooth ridges. If you put ridges in the inward facing tori shells, you have to check to see what they do to the magnetic field. With very strong fields such ridges may not have any effect at all.

There are three other hairy problems. One is feeding the liquid gasses to the core and innermost cooling jacket. Another one is tying the magnets together in a cubical array with the right gaps. The third is, what material do you use to do this job? I'm thinking something like Type 439 SS might work the best, as bizzarre as that sounds. Trouble is, I don't know how well that material will perform under the stringent vacuum requirements needed for this beast. I don't know that anyone has ever tested it. It may well outgass like a sick cow.

Have you sketched this thing yet?

[MSimon]

Dan,

Go to 2 m and you are down to 2 MW/ sq m.

And the fact that you are dealing with the shell of a torus. The area of the shell is going to be larger than the intercepted area.

[Aero]

larger by a factor of Pi?

[MSimon]

larger by a factor of Pi?

Pi over 2. It is a half shell. About 1.6 for most practical calculations.

[Billy Catringer]

I have been twisting this around in my head, not hard given the nature of my itty-bitty brain, and it looks to me as though the truncated dodecahedron is going to be the better of the two designs insofar as a stable structure goes. It is easier to make supporting triangles out of that design and I suspect the gapping problem will be much simpler to solve. Unfortunately, this design is a hell of a lot more expensive.

Also, I am thinking that rather than soft ridges running transverse to the long axis of the inward facing shells of the tori, it would be better to use sphircal humps with gently sloping shoulders. This should allow the metal to spring in two directions without causing trouble at the weld seams.

[Billy Catringer]

I just found a skecth on your blog, MSimon. Here's another possibility for you. First is the core surrounded by LHe. Next is a vacuum jacket. After than is a LN2 jacket followed by another vacuum jacket. Between this vacuum jacket and the first water jacket, add a jacket of ultra high molecular weight polyethylene, followed by another vacuum jacket and then a single large water jacket. With a coating of 10B on the outside of of the outermost shell, that should just about eliminate the neutron problem for your superconductors. Even better, the ultrahigh molecular weight polyethylene ring would add a great deal of strength to the entire assembly.

[tombo]

I estimate a multi minute run cooling Cu Will require pumping 400 l/m and will boil off 40 l/m

You don't make it. You order it by the truckload from the nearest dstr.

I checked into prices a while back. About $1 a liter in Dewar quantities. About $.75 in truckload quantities.

There is also a LHe plant in WI not too far from here.

i.e. 2400 l/hr
For trucks that makes $1800/hr as an upper limit price.
Assuming that volume was for wb100 that is 100MW.
[The context may have implied wb7 size. If so all bets are off.]
That gives $0.018 per KW-hr
comparing with $0.10 per KW-hr retail
That is an 18% economic drag.
Worse than that if the deliverable plant output is significantly less than the nominal 100MW. (Which some threads conclude.)
That is huge, perhaps even a show stopper if too many other things are non-optimal.

OTOH

A small LN2 plant consumes 13KW and produces 7L/Hr:
http://www.stirling.nl/sp/sp3.html

That would take 4.5 MW for cooling.
That is 4.5% of the power output which does not seem so bad.

OK any thermodynamcists out there?
Can we turn turbines effectively with the N2 that boils off in the hot zone?
That would be one way to recover some energy and partially re-cool the N2.
The turbine blades would live in a much more benign environment than those for superheated steam.
i.e. probably no fancy expensive materials, perhaps even plastic.
However the cold is a metallurgical problem to be solved in its own right.
I'm concerned about the low Carnot efficiency due to the low high-end temperature.
My TANSTAAFL alarm is going off. I'm afraid we will be stuck between a conflicting requirements.
But there IS 4.5MW of power there, some of which ought to be extractable.

(I apologize for the long delay in responding.)

[MSimon]

The context may have implied wb7 size. If so all bets are off.

Yes that was for a WB-7 size unit and the cooling was strictly for experimental purposes. Shot length seconds to minutes.

[Billy Catringer]

Vibration is going to be a problem as well. I think it will prove impossible to have all six of the magnets connected together in series in the coolant loop. That would require high velocity flow and the head on the discharge side of the coolant pumps would skyrocket.

Looking at the constraints imposed by the need to avoid disturbing the magnetic field lines, it will be necessary to try and feed two magnets to a loop from the the standoffs. Even that is going to be troublesome.

The LHe has to be fed through the higher temperature jackets, as will the LN2. These cryogenic feed lines will have to be insulated from the two water layers, both of which will be hot and at high pressure. Evaporative cooling here? If so, more plumbing that must be hidden from the magnetic fields.

Assuming that three cooling loops (including all four coolants in each loop), with two magnets in each loop, distribution becomes the next issue. At a minimum the system will require one LHE pump, one LN2 pump and two different water pumps because of the difference in temperature and pressure in the water jackets. That's four pumps minimum assuming this thing can be fed from from separate headers for each coolant. Then the system will need some pretty good compressors to deal with the gasses coming out of the system.

Maintaining the vacuum on the insulating jackets will likely not be such a terrible chore because the vacuums maintained in them need mot be of the same quality as the vacuum held around the outside of the magnets. Still, the system will need one powerful vacuum pump devoted to establishing and maintaining vacuum on those jackets.

Would chilled copper be good enough to drive a p-11B plant, or are the superconductors an absolute must?

[MSimon]

With copper the power losses are horrendous for the fields generated.

BTW MRI coils already handle two loops - LN2 and LHe. It will be a difficult but not impossible task to extend that to 4 loops. (2 water - 2 cryogenic).

[Billy Catringer]

I am one of those rare old farts that has yet to see an MRI! I've seen 'em on TV so I don't know if I am seeing all of one. Either way, we are talking about a monstrous great difference in scale, along with a huge difference in temperature differences and pressures in what amounts to a a single vessel, when we are describing a single magnet.

LHe is actually THE coolant in the design you propose, right? The two water loops are there primarily to provide a shield against neutrons, but they are doing double duty by convecting heat away from the deeper layers. LN2 is also an attempt to convect heat away before it reaches the LHe. The difference in temperature from the core to the outer shell is a doozy.

Don't get me wrong, I am not trying to be a doomsayer here. It's just that I have spent a lifetime trying to build and maintain the stuff real engineers dream up and this one is one of the real doozies.

If superconductors are an absolute must, then you need to try more than one approach to answer the cooling requirements. Keeping the two water jackets full of liquid rather than steam across every magnet in the system is going to demand the use of two great big pumps and the flow rates are bound to cause trouble with vibration, not from the pumps per se, but from the turbulent flow inside the torii.

[MSimon]

Don't get me wrong, I am not trying to be a doomsayer here. It's just that I have spent a lifetime trying to build and maintain the stuff real engineers dream up and this one is one of the real doozies.

This project is full of them. And that is just the engineering.