## 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

icarus
Posts: 819
Joined: Mon Jul 07, 2008 12:48 am
Scupperer you're asking the right question IMO.

Having the nubs located at the mid-point along the edge of the inscribing cube of the Polywell is a mistake I think. If you take a look at this approx. spherical wiffleball (scroll down for colourful "beachball" picture with lines on it), one of the lowest field points is directly beneath a nub (radially speaking).

viewtopic.php?t=650&postdays=0&postorder=asc&start=30

It seems counterintuitive that a field low-point, i.e. the mid-edge cusp, should occur beneath where the coils come closest together, but I think this spherical solution is saying something about the way the field changes when the wiffleball forms and pushes back the magneto-static field.

Anyway, having the nub radially above a cusp is like putting it right in the firing line for oscillating (recirculating electrons). Moving it to a point either side would seem to make sense, quarter-point? or maybe three nubs spaced at 120 degrees?

I think there is also a question raised here about why the spacing of the closest point between the coils should be kept to the minimum (a couple of gyro-radii I think Bussard said). If these mid-edge cusp form when the wiffleball forms anyway, why not make the spacing between the coils at the edges something like the diameter of the coils themselves? This will make the strength of the all the cusps about equal and close up the central point cusps, that will be the largest source of losses in the current configuration. The way I see it, the present the magnetic "cage" is not spreading material optimally over the surface of the sphere it is trying to cover, a significant portion of it is doubling up near those mid-points.

Torulf2
Posts: 285
Joined: Fri Sep 21, 2007 9:50 pm
Location: Swedem
Like this? But 3 support pips for every coil. The pipes are in shadow for the alpha radiation but the lean structure may make it more stabile.
http://farm4.static.flickr.com/3437/334 ... fd.jpg?v=0

icarus
Posts: 819
Joined: Mon Jul 07, 2008 12:48 am
Torulf2:

If you rotate the stand-off-coil connection 22.5 degrees, pi/8 radians about the coil center, it will be equidistant from the edge cusp and corner cusp, azimuthally speaking, and radially above a "high" field saddle location, (theoretically the least dense region of oscillating/recirculating electrons). However, since there are 8 of these high field points about the coil circumference you have a choice how many connections and which locations, somewhere between 3 and 8.

I like the idea of having the support in a radial path from the center, i.e. in the shadow of the coil. As a general rule, it's obvious the least material in the chamber besides the coils the better.

KitemanSA
Posts: 6114
Joined: Sun Sep 28, 2008 3:05 pm
Location: OlyPen WA
ekribbs wrote: Scupperer's idea here will put the semicircular "nubs" into massive bending due to the 60 Tesla magnetic forces you stated earlier, The "nubs" would break from the bending and the whole assembly would fly apart. 60 Tesla, WB-6 was 1.25 Tesla.
No. It was more like 1/10th that. It was 1.3kGauss, or 0.13T per EMC2's publication A QUICK HISTORY OF THE EMC2 POLYWELL IEF CONCEPT dated December 2006.

KitemanSA
Posts: 6114
Joined: Sun Sep 28, 2008 3:05 pm
Location: OlyPen WA
Scupperer wrote: I'm curious what effect reshaping the "nubs" as pictured above might have on the fields. Will changing their intersection to the rear of the coils, and increasing their radius clear them of the wiffleball effect and recirculating particles?
The nubs are only a problem where they are.

Stealing Indrek's beautiful rendition so ably found by icarus and sketching crudely on it:

the nubs are currently placed right in the middle of the dark blue oblongs which are the quasi-linear cusps that are on, but perpendicular to, the line between the really round cusps. Move the nub up and down a bit (a pair replace the one) and run the proper current thru them and they become the final part of the triangle's coil, and the color below them changes to orange like the main coil. That orange field protects the metal, so no losses, and the nubs can provide all the load carrying capacity you need, without introducing odd torques or moments into the system. At that point, the need for standoffs is minimized. I think we can do just fine with the same number, 4, as the WB6, but I think they should suspend, not support the MaGrid.

KitemanSA
Posts: 6114
Joined: Sun Sep 28, 2008 3:05 pm
Location: OlyPen WA
icarus wrote:It seems counterintuitive that a field low-point, i.e. the mid-edge cusp, should occur beneath where the coils come closest together, but I think this spherical solution is saying something about the way the field changes when the wiffleball forms and pushes back the magneto-static field.
Actually, I don't think it is a low point per-se, but a point with very low lateral vector. The field is moving mostly into, or out of that sphere at the blue, and mostly tangential at the orange. Tangectial (orange) reflects, radial (blue) transmits. Only where there is a funny cusp (pairs of opposite polarity fields meeting) might there be an actual field null.

KitemanSA
Posts: 6114
Joined: Sun Sep 28, 2008 3:05 pm
Location: OlyPen WA
ekribbs wrote:2) If you get a picture from online somewhere, your semicircular tubes STILL cross magnetic lines of force and provide an electron loss path. The only components that may be aligned to lie only in a "shadow" would be the base legs! I also stated that all the truss components must be insulated as well.
Folks, there is no shadow to the electrons. Those that escape a cusp will follow lines of magnetic field in a curve until they either stop, reverse, and return to the cusp from wence they came; or they zip around the line to the next cusp; or they run into one of your excess legs. The shadow may apply to the ~3MeV alphas, but NOT to the electrons.
Minimize the matter in the chamber, and magnetically protect all you can.
Nubs place slightly away from the corners and carrying current does this.
Of course I don't really want "nubs", I want the triangular coil to become real, not virtual.

KitemanSA
Posts: 6114
Joined: Sun Sep 28, 2008 3:05 pm
Location: OlyPen WA
Scupperer wrote:
ekribbs wrote:You must remove the semi-circular loops because:
1) radial outward forces will break them.
2) they provide an electron loss path.
1) Not if they're not the primary supports and are just acting in a lateral support capacity, as shown in the previous post.
Even for lateral support, the horse-shoe seems way too flexible. Yeah, I suppose it is just a matter of enough matter, but the dual, magnetically protected new-nubs (triangle-coils) fulfill the need without ANY metal in the way (I think).
Scupperer wrote:2) This is really the only concern. I'm curious if this is the case, because the idea of changing the shape is to start in the shadow on the rear of the coil and go around the cusp line, instead of directly through it.
You can't go "aound" the cusp line. The cusp line is the intersection of radial field lines and the nominal sphere. The field that exits the sphere along that cusp line continues out a LONG way. (Inward, it is deflected out by electron pressure).
Scupperer wrote:I'll admit my caveman understanding here, though, yet I shall forever wonder why the supports themselves aren't an electron loss path, too.
IMHO, they are. See my prior post. New mantra, "the electrons got no shadow".

ekribbs
Posts: 13
Joined: Thu Feb 26, 2009 3:04 pm
Location: Laurel, MD
KitemanSA wrote:
Scupperer wrote: I'm curious what effect reshaping the "nubs" as pictured above might have on the fields. Will changing their intersection to the rear of the coils, and increasing their radius clear them of the wiffleball effect and recirculating particles?
The nubs are only a problem where they are.

Stealing Indrek's beautiful rendition so ably found by icarus and sketching crudely on it:

the nubs are currently placed right in the middle of the dark blue oblongs which are the quasi-linear cusps that are on, but perpendicular to, the line between the really round cusps. Move the nub up and down a bit (a pair replace the one) and run the proper current thru them and they become the final part of the triangle's coil, and the color below them changes to orange like the main coil. That orange field protects the metal, so no losses, and the nubs can provide all the load carrying capacity you need, without introducing odd torques or moments into the system. At that point, the need for standoffs is minimized. I think we can do just fine with the same number, 4, as the WB6, but I think they should suspend, not support the MaGrid.
The designer had to bump the major Radius to 2 meters to accommodate all the multiple casings for the various stages of coolants to make the magnet super-conduct. With all the concentric casings, wirings, mass of coolant, etc., the weight of just one toroidal magrid came out to 60,000 Kg (132,000 lbs.). Total weight so far is 792,000 lbs for 6. I would like to be there when, after construction is finished and construction supports removed, to watch the little nubs crumple and the whole structure collapse down onto the floor.

The designer also says that he cannot get even cooling or even adequate coolant through those nubs for the magnets, or cool the nubs either. The nubs have to go. Each torus is going to have to be cooled separately with separate piping. That extra piping may as well be structural.

Why doesn't anyone complain about electron losses in the 4 base supports from WB-6?
Because they are insulated. And so I also stated that the new supports must be also insulated.

MSimon
Posts: 14333
Joined: Mon Jul 16, 2007 7:37 pm
Location: Rockford, Illinois
Contact:
Scupperer wrote:
ekribbs wrote:You must remove the semi-circular loops because:
1) radial outward forces will break them.
2) they provide an electron loss path.
1) Not if they're not the primary supports and are just acting in a lateral support capacity, as shown in the previous post.
2) This is really the only concern. I'm curious if this is the case, because the idea of changing the shape is to start in the shadow on the rear of the coil and go around the cusp line, instead of directly through it. I'll admit my caveman understanding here, though, yet I shall forever wonder why the supports themselves aren't an electron loss path, too.
Any thing near the coils not in their shadow is going to have to deal with a heat flux of around 1 MW/sq m. It will have to be cooled.

The supports are not an electron loss path if they are in the shadow of the coils. Or mostly in the shadow of the coils. Why is that? Because it is a beam machine (mostly).

May I suggest a thorough reading of the Valencia Paper to get some idea of the constraints.
Engineering is the art of making what you want from what you can get at a profit.

icarus
Posts: 819
Joined: Mon Jul 07, 2008 12:48 am
Hmm, after looking at the below diagram some more I'm having second thoughts about the contention that nubs directly above mid-edge cusp is bad. The osculating coils, that are not shown in this section since we are looking at the diagonal section that bisects them, are at about a similar radius as the large black coils shown. At that level, the field lines are quite uniformly radial and the slight orange tinge suggests the field is stronger at that level directly above the mid-edge cusps.... more head scratching is in order, I think.

Billy Catringer
Posts: 221
Joined: Mon Feb 02, 2009 2:32 pm
Location: Texas
Torulf2 wrote:Like this? But 3 support pips for every coil. The pipes are in shadow for the alpha radiation but the lean structure may make it more stabile.
http://farm4.static.flickr.com/3437/334 ... fd.jpg?v=0

That's pretty close insofar as the general configuration goes, Torulf.

There are some questions I still need answers to.

1) How much room doest this physics require around the magrid?

2) How far away from the magrid must the alpha catcher be from the center of the magrid?

3) How big is the alpha catcher going to be, especially in depth?

We will have to build a vacuum vessel around all this stuff and, so far, I haven't a clue as to what its diameter might be.

tombo
Posts: 334
Joined: Fri May 09, 2008 1:10 am
Location: Washington USA
Torulf,
Good, but where do you put the 3rd leg for symmetry? (on a coil next to 4 others)
I’ve been scratching my head over this one for a while too.
I added the 4th one because I really like symmetry.
One could go for 8 triangular coils with 3 legs each instead of 6 square coils with 4 legs each, but the leg count is 24 either way and 6 is fewer coils to build and they each have 90 degree bends which seem easier than 120 degree bends.
But with 4 legs I could all too easily see it twisting, racking and wobbling like a pendulum. So, I doubled the number to give triangle strength and resist buckling initiation loads. It looks like a lot of legs, but there are huge loads that we need to resist.
I like the truss shape that I came up with by following that path. I hope you can find a better solution.

Kiteman:
I agree it will have an X shaped region of zero field.
I disagree that zero fields are fine.
DrB never drew it this way to my knowledge. [Although, I vaguely remember that one of the enclosed machines had vertex coils and did not work very well for reasons that Art Carlson was very eloquent and adamant about a few months back.]
When I look at it I see 48 vertexes, each with an odd number (3) of faces.

Of interest there are irreconcilable geometrical issues with the shape.
If the hole is square then the straight to curved transitions cannot be in the same place for the two curves. If they are the same then the hole is very much longer than it is wide. I tried to make that work for hours.

Without the crossovers there are no zero field points:

I believe that DrB referred to square plan form, round cross section coils like these. (Although maybe closer to touching than shown in this plot.)
IMO the funny cusp is an artifact of approximating the ideal infinitely sharp vertex points with the long rounded almost kissing regions of round coils like in WB6.
I further believe that the square plan form reduces it greatly. The sharper the bends can be made, the smaller the funny cusp becomes. It reduces in size to approximately what it would be in a machine with round coils the same radius and spacing as the radius of the curved corners of the square coils.

Here without my notes for your marking pleasure: http://i299.photobucket.com/albums/mm31 ... sOver2.jpg

P.S. Here it is up close:
http://i299.photobucket.com/albums/mm31 ... sOver4.jpg
-Tom Boydston-
"If we knew what we were doing, it wouldn’t be called research, would it?" ~Albert Einstein

Billy Catringer
Posts: 221
Joined: Mon Feb 02, 2009 2:32 pm
Location: Texas
I made this partial model to double check the dimensions I calculated for the magnet placement. This, of course, is the Belarus Tractor version of the magrid:

There is a 2cm clearance between the outermost jackets on the solenoids. You can see the faint crack in the foreground of this render. The distance from the center of the magrid to the center of the SC Coils shown in this picture is 295.85 cm. The corner-to-corner dimension of the magrid (based on the outside diameter of the Hot Water Jackets) is 836.79cm or, if you prefer English Units, 27' 5 1/2". I left out all of the internal jackets, in case you were wondering. I did not need them for this particular exercise.

This is about as close as I can get the magnets together with this cooling system. I will do some more work on it in an effort to reduce its size and weight.

KitemanSA
Posts: 6114
Joined: Sun Sep 28, 2008 3:05 pm
Location: OlyPen WA
Billy Catringer wrote:This is about as close as I can get the magnets together with this cooling system. I will do some more work on it in an effort to reduce its size and weight.
You may consider trying LO2 rather than LN, and Fluorinert rather than the inner water course. they both have significanly greater liuid temperature ranges which lowers the pumping and volume requirements. Also, LO2 has a lower melting point than LN, so it can be chilled more. This will reduce the thermal load to the LHe circuit markedly. Along that vein, a ~20%N80%O eutectic has an even lower freezing point. Something to think about.
Second, the 1+cm walls an all those courses look WAY excessive unless you are trying to restrain boiling (see first comment).
Third, I don't see any real thermal radiation shielding, but that may just be because it is hard to depict graphically. What I mean here is multiple very thin sheets of foil.
Finally, look up truss core. This is a very light weight, pyramidal truss core panel that makes an excellent and extremely light weight heat exchanger. It is also reasonable easy to manufacture in tubular shapes using metal glass brazing.
I suspect you might be able to get the heat control system much smaller than it is now. For one thing, you are working with evacuated systems in a vacuum chamber. Not a lot of external, buckling type, pressure loads.