Is There an Optimal Size for Magrid Casings?

Discuss the technical details of an "open source" community-driven design of a polywell reactor.

Moderators: tonybarry, MSimon

charliem
Posts: 164
Joined: Wed May 28, 2008 8:55 pm

Is There an Optimal Size for Magrid Casings?

Postby charliem » Fri Jun 20, 2008 2:08 am

MSimon wrote:
dch24 wrote:20% of a 100MW fusion reaction is a LOT of power barraging the MaGrid.


We have a solution: a LOT of heat transfer.

About 100 to 400 gpm of cooling water per grid. Not too bad. (multiply by 4X to get l/min - close enough for now).


Simon, I think you have some BOE figures about the dimensions of the cooling system for a given temperature differential, let's say 523 K external and 4 K internal (LHe), or 523 K external and 77 K internal (LN2), could you comment on that?

I'm thinking that building a magrid that only intercepts a 20% of the alphas might prove difficult given the added bulk of the cooling (specially if it is multi-layer), and also that more distance between coil and shell means a lessened superficial magnetic field.

I'd like to explore the possibility of using non SC magnets at a higher temperature (less layers of cooling). I have the equations for heat transfer, but those dont tell you anything about real world considerations (that is about engineering), and without that I cant proceed.

It might well be that there is an optimal configuration that is not evident.

MSimon
Posts: 14331
Joined: Mon Jul 16, 2007 7:37 pm
Location: Rockford, Illinois
Contact:

Postby MSimon » Fri Jun 20, 2008 2:28 am

Charlie,

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

For LN2 cooled Cu magnets I think I allowed 10K delta T. Maybe 20K. LN2 can't be liquid at any temp above 120K. It boils at 77K. You can get some help by boiling at a lower pressure. 70K is probably the lower limit with 72K safer (from solidification).

I don't see how you get around multilayer cooling. An outer layer to carry away the heat. A buffer layer. A LN2 layer. A LHe layer at the center. With vacuum between each layer.
Engineering is the art of making what you want from what you can get at a profit.

charliem
Posts: 164
Joined: Wed May 28, 2008 8:55 pm

Postby charliem » Fri Jun 20, 2008 3:46 am

MSimon wrote:Charlie,

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

For LN2 cooled Cu magnets I think I allowed 10K delta T. Maybe 20K. LN2 can't be liquid at any temp above 120K. It boils at 77K. You can get some help by boiling at a lower pressure. 70K is probably the lower limit with 72K safer (from solidification).

I don't see how you get around multilayer cooling. An outer layer to carry away the heat. A buffer layer. A LN2 layer. A LHe layer at the center. With vacuum between each layer.

Lets see if my numbers are right.

If magrid cassing is a torus of 1.8 m (medium diameter) x 20 cm (diameter of the projected circle), its total surface would be 14'2 m2 (a little less in the inside) and it'd intercept about 4'7 MW in alphas from a total of 100 MW (aprox 5% each coil).

Heat load would be about 0.35 MW/m2 for the first layer.

Water flow needed to keep delta T = 100 C is 47 l/s.

What I dont know how to calculate is the thickness of the water blanket (only have equations for a round section pipe) to keep a given in-out pressure drop (I think you said something about 100 psi or aprox 7 atm).

And how thick is the double walled vacuum separation between layers? I can calc the radiated heat interchange of two surfaces at given temperatures, but what about structural integrity, and what about conduction through supports separating them.

We need coils that not just give a certain B at their centers, but also have a very steep gradient near their casings, and that means that the layered cooling has to be as thin as possible.

Will be my idea of switching the torus for a conical frustum compatible with multilayered cooling? :?

Well, I'm going to muse on it for a while and see if I can come with anything useful.

MSimon
Posts: 14331
Joined: Mon Jul 16, 2007 7:37 pm
Location: Rockford, Illinois
Contact:

Postby MSimon » Fri Jun 20, 2008 4:07 am

I don't think the vacuum needs to be very thick at all. As long as it is a conductivity barrier the radiation will take care of itself. A silvering (alumanizing?) of the vacuum sides of the plumbing might help. A lot will depend on things like creep and hoop stress, and temp, and temp delta T in the loop etc. and mfg tolerances.
Engineering is the art of making what you want from what you can get at a profit.

nferguso
Posts: 55
Joined: Sat May 17, 2008 3:43 am

Postby nferguso » Sat Jun 21, 2008 1:54 pm

The perspective of discussion I have seen is in reference to minimizing the first wall surface area. But how about taking the opposite tack? Embrace the hot, I say! Make the magnet housings as big as the fluxes can allow. Surface area goes up with the square, but coolant volume goes up with the cube. You have more room for mechanical rigidity. Bigger might mean more congenial magnet constraints.

Go with steam turbines as your major energy output. Worry about the mega-voltage later. KISA. If break-even is reached, platoons of engineers will optimize the design for applications.

MSimon
Posts: 14331
Joined: Mon Jul 16, 2007 7:37 pm
Location: Rockford, Illinois
Contact:

Postby MSimon » Sat Jun 21, 2008 8:14 pm

nferguso wrote:The perspective of discussion I have seen is in reference to minimizing the first wall surface area. But how about taking the opposite tack? Embrace the hot, I say! Make the magnet housings as big as the fluxes can allow. Surface area goes up with the square, but coolant volume goes up with the cube. You have more room for mechanical rigidity. Bigger might mean more congenial magnet constraints.

Go with steam turbines as your major energy output. Worry about the mega-voltage later. KISA. If break-even is reached, platoons of engineers will optimize the design for applications.


Heat transfer is an area problem. It doesn't matter how much water you pump through a pipe. What matters is how much contacts the surface. Turbulent flow is the thing. High surface to volume ratios help keep the Reynolds number up.

Steam turbines are high cost long lead time items. They are also low efficiency. Direct conversion is the way to go. Use the heat from the grids for industrial processes.
Engineering is the art of making what you want from what you can get at a profit.

TallDave
Posts: 3114
Joined: Wed Jul 25, 2007 7:12 pm
Contact:

Postby TallDave » Wed Jun 25, 2008 3:12 am

I don't think the vacuum needs to be very thick at all. As long as it is a conductivity barrier the radiation will take care of itself. A silvering (alumanizing?) of the vacuum sides of the plumbing might help. A lot will depend on things like creep and hoop stress, and temp, and temp delta T in the loop etc. and mfg tolerances.


I'd give my left arm to see those detailed reactor designs Dr. Nebel mentioned the other day.

charliem
Posts: 164
Joined: Wed May 28, 2008 8:55 pm

Postby charliem » Wed Jun 25, 2008 5:49 pm

TallDave wrote:I'd give my left arm to see those detailed reactor designs Dr. Nebel mentioned the other day.

You're not the only one. :D
BTW, I hope that Dr. Carlson accepts your invitation to come.

And talking about something else, it seems that making a WB100 magrid with copper at high temp (233K and up) is doable. It'd have some advantages, namely the reduced size and complexity of the cooling system and the increased ratio 'cusp field to coil center field'.

But there is a drawback, heat generated inside the coils by Joule effect would be of the same order of magnitude than heat received from fusion (megawatts). To sidestep that we could cool the coils down to LN2 temperature (77K), or even better to LHe (4K), but then why using copper at all?

tombo
Posts: 334
Joined: Fri May 09, 2008 1:10 am
Location: Washington USA

Postby tombo » Wed Jun 25, 2008 7:02 pm

Has anyone run the power numbers for refrigerating that much LN2?
-Tom Boydston-
"If we knew what we were doing, it wouldn’t be called research, would it?" ~Albert Einstein

MSimon
Posts: 14331
Joined: Mon Jul 16, 2007 7:37 pm
Location: Rockford, Illinois
Contact:

Postby MSimon » Wed Jun 25, 2008 7:40 pm

But there is a drawback, heat generated inside the coils by Joule effect would be of the same order of magnitude than heat received from fusion (megawatts). To sidestep that we could cool the coils down to LN2 temperature (77K), or even better to LHe (4K), but then why using copper at all?


My interest in copper coils was strictly for experimental purposes with runs on the order of seconds to tens of minutes. In such a situation LN2 cooling makes sense.

The reason for Cu was that such a design could be done quickly vs superconductors.
Engineering is the art of making what you want from what you can get at a profit.

JohnP
Posts: 296
Joined: Mon Jul 09, 2007 3:29 am
Location: Chicago

Postby JohnP » Wed Jun 25, 2008 11:05 pm

tombo wrote:Has anyone run the power numbers for refrigerating that much LN2?


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

MSimon
Posts: 14331
Joined: Mon Jul 16, 2007 7:37 pm
Location: Rockford, Illinois
Contact:

Postby MSimon » Thu Jun 26, 2008 1:24 am

JohnP wrote:
tombo wrote:Has anyone run the power numbers for refrigerating that much LN2?


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


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.
Engineering is the art of making what you want from what you can get at a profit.

MSimon
Posts: 14331
Joined: Mon Jul 16, 2007 7:37 pm
Location: Rockford, Illinois
Contact:

Postby MSimon » Thu Jun 26, 2008 1:50 am

I estimate a multi minute run cooling Cu Will require pumping 400 l/m and will boil off 40 l/m meaning that a 1,000 liter total supply should be good for 10 minutes. With 50% left over to keep the pipes full. And 5 hours to make up the loss for the next run.

Using the 88 l/ hr machine and the same size WB as is currently testing (.3 m dia coils) with Bitter type Cu magnets.

The thing to do is to do a few test runs with trucked LN2 and then if usage was going to be high (i.e. machine uptime good) buy an LN2 maker.

Do a long run in the morning (16 hours of liquefier run) and then a shorter run near the end of the day. Or 2 minutes of run time every hour.
Engineering is the art of making what you want from what you can get at a profit.

charliem
Posts: 164
Joined: Wed May 28, 2008 8:55 pm

Postby charliem » Thu Jun 26, 2008 2:32 am

Or even simpler. If they ever get to commercialize those WB7-like machines Dr Nebel talked about I hope he can add them a little pre-cooling at least.

WB-6, starting from ambient temp, produced non constant fields of 0.1-0.15 T for 20-30 secs before overheating.

With a constant intensity power source and starting from 77K (LN2) it could double the field or quadruple the time, and precooled to 4K the coils would generate so little heat that it could work at 0.1-0.15 T for half an hour or more, or try Bs over 0.6 T (supposed the structure could bear the forces).

ravingdave
Posts: 650
Joined: Wed Jun 27, 2007 2:41 am

Postby ravingdave » Thu Jun 26, 2008 7:50 pm

MSimon wrote:
nferguso wrote:The perspective of discussion I have seen is in reference to minimizing the first wall surface area. But how about taking the opposite tack? Embrace the hot, I say! Make the magnet housings as big as the fluxes can allow. Surface area goes up with the square, but coolant volume goes up with the cube. You have more room for mechanical rigidity. Bigger might mean more congenial magnet constraints.

Go with steam turbines as your major energy output. Worry about the mega-voltage later. KISA. If break-even is reached, platoons of engineers will optimize the design for applications.


Heat transfer is an area problem. It doesn't matter how much water you pump through a pipe. What matters is how much contacts the surface. Turbulent flow is the thing. High surface to volume ratios help keep the Reynolds number up.




You probably already thought of this, but it might be of interest to others. Does everyone remeber those old pith ball experiments where a pith ball is suspended between two charged plates ? The pith ball bounces back and forth rapidly between the plates exchanging charges.

Well, from what i've been reading over the past couple of years, this concept also works for air, and at least one researcher was experimenting with an ion cooling system for microprocessors. Because of a high voltage charge on the plates, the air molecules are brought into intimate contact with the suface, and therefore the heat transfer per molecule is maxed out. With air flowing over the surface in a conventional cooling system, air forms a boundary layer next to the surface that prevents efficient heat transfer.

The Ion air cooling system punches through the boundary layer and maximizes air contact with the surface. From what I remeber, the energy required to cool something using this technique is dramatically lower than using a cooling fan.


What am I getting at? If this system works for air, it seems to me the same concept should work for a non conductive liquid.

Furthermore, if 20% of your alphas are striking the MagGrid, all that is required to achieve a potential difference between the outside wall of the grid and the surface of the next layer is to control the bleedoff of voltage from the outside of the MagGrid. Maintain the Inner layer at a lower potential, and the coolant molecules should ping back and forth between inner and outer layer as it flows through the system.

Talk about your turbulence ! I don't think you can get more turbulent than that !

Just an observation. Might be wrong. Dunno.


David


Return to “Design”

Who is online

Users browsing this forum: No registered users and 3 guests