Thermal modeling -- request for help

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

Moderators: tonybarry, MSimon

Post Reply
FAMULUS
Posts: 32
Joined: Tue Dec 23, 2008 12:02 am
Contact:

Thermal modeling -- request for help

Post by FAMULUS »

The fusor's red hot grid reminds me I can't ignore heating from ion bombardment as we design a superconducting polywell. This won't cut it:

http://prometheusfusionperfection.com/2 ... agrid-irl/

The inner core of the magrid must stay at liquid nitrogen temperatures, while the exterior skin of the magrid may get very hot from ion bombardment (from both plasma and fusion).

I will try to implement MSimon's idea detailed here:

http://iecfusiontech.blogspot.com/2008/ ... oling.html

Although ours would start at the liquid nitrogen level.

I want to do computer thermal modeling on the candidate designs.

Here is where I NEED HELP:

1) I need to connect with someone who has thermal modeling experience

2) I need to source and learn software to build thermal modeling into the design workflow.


There is software out there:
http://www.thermalsoftware.com/

So far I have not found any open source package for thermal modeling.

Most important is talking to someone with experience doing it.

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

Post by MSimon »

Thermal modeling is iffy. It is more helpful with: "if I change this what direction will that go in". Absolute values are good for getting you in the ballpark.

First thing to do is some calorimetry. Find out what your actual (no magnet - probably worst case) heat load is. DI water is probably good if you can dry the magnets before using LN2 to cool them.

Of course you will have better ideas being more intimate with your set-up. But finding the actual heat load is a good place to start.

Another place to start is to assume all your power supply power has to be dissipated.

Find out what the thermal conductivity of your outer shell is. Maximum temperature desired. Calculate the flow. Make designed flow 2X or 3X calculated flow to account for hot spots and design imperfections. Flow should be turbulent. So a Reynolds number for your setup would be good.

If you are going with a custom shell the side in contact with the water should be as rough as possible. The thickness of the water channel should be minimum to keep the flow turbulent. All in keeping with the maximum pressure possible (including some "overpressure" to keep the water from bulk boiling inside the reactor. That may mean a throttling valve on the outlet from the reactor.

Such rules of thumb should be adequate for anything less than a full up commercial design.
Engineering is the art of making what you want from what you can get at a profit.

Axil
Posts: 935
Joined: Fri Jan 02, 2009 6:34 am

Post by Axil »

Make the whole reactor out of diamond; it is cheap and will both conduct of insolate based on doping. It does not interfere with magnetism.

Surround the reactor with a FilBe blanket. Run it at 700C for power removal and shielding (700C is the ideal diamond defect annulling temperature).

This blanket will not see magnetism.

Make the blanket shell a double walled thermos bottle with a vacuum between the walls to keep the blanket heat from the magnet.

Place the magnet outside all. Do you still think that this won’t work?

Search SourceForge for "thermal", you will find the software you want.

http://sourceforge.net/

D Tibbets
Posts: 2775
Joined: Thu Jun 26, 2008 6:52 am

Post by D Tibbets »

Actually, I do not think you have to worry about ion bombardment much. If your magnetic shielding is working , only rare ions should reach the magnets. The nubs/ interconnects between coils might intercept some ions, though I think electrons are hitting the nubs much more often.
My impression of the heating of the magrids in order of magnitude of a tabletop Polywell at perhaps 10-20,ooo volts would be:

1) Ohmic heatig from current in the copper magnet wires (substantially less if at liquid nitrogen temperatures).

2) X-ray heating from bremsstrulung and electron- wall impacts.

3) Electron heating of the nubs. The total flux may be modest, but it is concentrated on the small nubs so might be a local heating problem.

I speculate that the magrid in the vacuum will be insulated well enough without additional walls. The amount of liquid nitrogen flow to overcome the wire heating will probably only need slightly increased flow to overcome external heating. The nubs could be protected with a small standoffs formed of malleable sheet metal. The sandcastings may need to be polished to prevent arcing.


The heating from the copper windings at liquid nitrogen temperatures and the liquid nitrgen flow rates needed to maintain the temperature should be straight forward (for an engeener). The external loads should be small. The electron current devided fraction that you guess will eventually hit the nubs should give the electron heating. Add a fudge factor for shell heating from escaped electrons, ion, and also the guess of how many high speed neutrals that may be bouncing around. Then figure how much this will heat the magrids through radiation and conduction through the supports, etc.

Any fusion heating from neutrons (and any charged particles) will be well under a few milliwatts, unless you are trying to substantially surpass the performance of WB6, in which case carfull concideration of radiation exposure, both x-ray and neutrons, needs to be concidered.

Again, I expect all these heat sources will be minor compared to the Ohmic heating of the magnet wires, even at liquid nitrogen temperatures.

Dan Tibbets
To error is human... and I'm very human.

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

Post by MSimon »

D Tibbets,

He will be lucky to get .1 T - not good enough for magnetic shielding.

He is using YBCO SC wire (tape) to carry the current.
Engineering is the art of making what you want from what you can get at a profit.

D Tibbets
Posts: 2775
Joined: Thu Jun 26, 2008 6:52 am

Post by D Tibbets »

MSimon wrote:D Tibbets,

He will be lucky to get .1 T - not good enough for magnetic shielding.

He is using YBCO SC wire (tape) to carry the current.
0.1 T is not enough to shield the magnets? That is what WB6 used, and presumably formed a wiffleball and thereby was shielded from the electrons. I'm too lazy to calculate the ion gyro radius at 10,000 volts speeds, but with the potential well the vast majority of fuel ions should be traveling slowely enough at the Wiffleball border that most ions upscattered enough to escape the potential well will be turned by a 0.1 Tesla field (?), except at the cusps, and these ions escaping the cusps shouldn't hit the magrid, but only the vacuum vessel walls and possibly some of the support structure (like the nubs. Since the nubs/ interconnects are on the inner half of the magrids, they may be hit more by the electrons, compared to the WB6, or WB7. Certainly any fusion neutrons or fusion ions at MeV velocities will hit the grids, but this should only add a few milliwatts of heating at most. I believe that WB6 produced a little less than one milliwatt of fusion energy (and that is only with extrapolation to running steady state for one second).

I speculate that the heat loads from fusion, even in WB8 will be very modest. A few watts (refrigerater light) assuming it runs for a full second or more. And that is total output, only a portion will hit the grids. The heat load from cooled copper wires will swamp this. a superconductor would theoretically produce no heat load.

At least that is my theoretical argument. Concidering that WB6 had about 40 amps of electron current when it was passing through Beta=1, and ~ 12,000 drive volts, the watts from the electron flow would be ~ 480,000 watts. . So even if the nubs only intercepted 1 % of the electrons and the rest of the magrid was perfectly shielded, there would be ~ 4,800 watts of heating of the nubs. The rest of the electrons would be hitting the vessel wall (ignoring some transport to the magnet surfaces. Mmm.. There would be significant heating of the nubs in a steady state machine. But in a pulsed machine that only ran for ~ 1 millisecond, the heating would onley add up to ~ 4.8 watts. That would not heat the mass of metal much in that short time frame.
In WB6 the copper coils (two hundred turns) and a couple thousand amps traveling through it (at 12 volts?). I don't know how much heating that caused over the 1 second or less time that the magnets were turned on, but I'm guessing it was conciderable. If a superconducter is being used, this heat will be absent.
If he plans to only do pulsed tests lasting ~ 1 ms. I'm guessing the liquid nitrogen flow will only need to be enough to compensate for heat conduction through the support structures. The magrid itself will be well insulated by the chamber vacuum. If he plans for longer runs, I suspect the cooling requirements will rapidly increase- mostly to keep the nubs cool, and to compensate for the heat conduction through the support structures, and radiated from the shell.
Of course, if he is using a few turns of superconducting ribbon and one or several turns of the ribbon through the nubs, they will be magnetically shielded to a significant degree and should experiance much less electron impact heating.

This all assumes he will be using ion guns or much more efficient and percise gas puffers than in WB6, to maintain a very low neutral population (in a steady state machine).

[EDIT] X-ray heating would be a portion of the heat load, but could not be more than a fraction of the total electron power. A larger percentage of this radiated 'heat' might be delivered to the magrid without needing to be conduct through the structure (standoffs) which probably would be good heat insulators such as ceramics.

Dan Tibbets
To error is human... and I'm very human.

FAMULUS
Posts: 32
Joined: Tue Dec 23, 2008 12:02 am
Contact:

Post by FAMULUS »

D Tibbets,

I'm using superconducting cable for this design, so no coil heating.

FAMULUS
Posts: 32
Joined: Tue Dec 23, 2008 12:02 am
Contact:

Post by FAMULUS »

Here is a rough draft of a superconducting magrid with a vacuum insulated heat shield.



http://prometheusfusionperfection.com/2 ... at-shield/

D Tibbets
Posts: 2775
Joined: Thu Jun 26, 2008 6:52 am

Post by D Tibbets »

At first glance a vacuum blanket would seem to be a good idea. But to confuse others perhaps as much as I confuse myself, concider the following. The magrid already has a spaced vacuum blANKET. it is the vacuum chamber wall. In an expermantal system in which the fusion products, wheather neutrons or charged particles are at most a few mllijueles per second (milliwatts), it can be ignored. For short runs (milliseconds) it is completely irrelivant. The local magrid outer wall would absorb any incident radiation just like the vacuum vessel walls. But these walls are very close to the inner wall of the magrid so sny heat would be radiated at a faster rate than from the vessel walls. Also, they would not benifit from as much heat conduction supression that would be provided by the long ceramic standoffs (or to a lesser extent metal standoffs) that support the magrid. The insulating blanket may help some , but I suspect the benifit is much less significant than you night think, at least in conditions where fusion power is trivial. The extra wall may decrease the x-ray dose that is absorbed by the superconductor, but this is a funtion of the increased mass of material , not an intervening vacuum. The thermal mass of the magrid metal casing in total may be the the limiting factor in terms of how much coolent flow for a given time is needed. An intervening vacuum blanket would help some, but I am speculating that in this application it would be next to insignificant.

Dan Tibbets
To error is human... and I'm very human.

Post Reply