Heat exchangers

[Darky]

I've read about the problem of withdrawing excess thermal energy from superconducting coils and remembered about another project that already developed similar tech, if I am not mistaken.

It's Skylon SSTO project. They were in the need of heat exchangers to rapidly cool incoming hot air from 1000C to -140C for their SABRE engine to work. So they were able to develop required technology. I quote:

The precooler is also the most aggressive and difficult part of the whole SABRE design. The mass of this heat exchanger is an order of magnitude better than has been achieved previously; however, experimental work has proved that this can be achieved. The experimental heat exchanger has achieved heat exchange of almost 1 GW/m³, believed to be a world record. Small sections of a real precooler now exist.

These heat exchangers have a mass of 1250 kg and are designed to transfer about 400 MW of heat at Mach 5.

So, why not using their solutions for Polywell?

Links:
http://en.wikipedia.org/wiki/SABRE
http://en.wikipedia.org/wiki/Skylon
http://www.reactionengines.co.uk/ - company site.
They also answer emails, as I already wrote them asking some questions about their project.

[Roger]

At mach 5 ... thats a large volume of air.... If we could get LN2 thru a magnet at that level the open source WB-7x might run for hours.

[TallDave]

Interesting, but probably not transferable to Polywell:

The 'hot' helium from the air precooler, and cooling the combustion chambers is recycled by cooling it in a heat exchanger with the liquid hydrogen fuel.

The loop forms a self starting Brayton cycle engine, and is used to both cool critical parts of the engine, but also to power turbines and numerous miscellaneous parts of the engine.

The heat passes from the air into the helium. This heat energy is not entirely wasted, it is in fact used to power the various parts of the engine, and the remainder is used to vapourise hydrogen, which is burnt in ramjets.

Besides the mechanics, that sounds very very expensive.

[Roger]

Interesting, but probably not transferable to Polywell:

Something about the mach 5 flow makes me think the same thing.

[MSimon]

Interesting, but probably not transferable to Polywell:

The 'hot' helium from the air precooler, and cooling the combustion chambers is recycled by cooling it in a heat exchanger with the liquid hydrogen fuel.

The loop forms a self starting Brayton cycle engine, and is used to both cool critical parts of the engine, but also to power turbines and numerous miscellaneous parts of the engine.

The heat passes from the air into the helium. This heat energy is not entirely wasted, it is in fact used to power the various parts of the engine, and the remainder is used to vapourise hydrogen, which is burnt in ramjets.

Besides the mechanics, that sounds very very expensive.

Suppose H2O is the final cooling mechanism vs LH2? Suppose under such a design you only get 100 MW/m^2. That takes Polywell from 100 MWth to 10 GWth with room for growth. And if LH2 is used as reaction mass for a rocket then we have a viable SSTO vehicle. Especially if a maglev track is used for initial boost.

[Darky]

Well, as I said you can write them and ask if their heat exchangers can be used in the way you want. And yes, it will be probably complex and expensive - but as well be the cheapest part of working fusion reactor!

The mach 5 number mentioned, I believe, means that the heat exchangers can cool up to M5 airflow speed, not they actually need supersonic airflow to work.

[Brent]

I posting this in hopes that it might be useful. Having not looked into the coils too deeply, and their heat transfer requirements, I cannot say it will be much of a help.

Check out microchannel heat exchangers. The paper that I have provided below states that heat transfer is higher in small channels than large ones.

See:
Investigation of heat transfer in rectangular microchannels

This is good if you want to get a lot of heat transfer for a small volume.
In addition, if the flow rates are low, pressure drop won't be as much of an issue.

Here is a second paper that is more accessible, but instead it talks about pressure drop in microtubes. It may at least give some idea of what kind of system a microchannel heat exchanger is.

And finally, here is a third paper that talks about heat transfer in microchannel heat exchangers. If you can't read it, at least the abstract is interesting. Notice the 300 W/cm^2 objective.

There is also a short blurb about microchannel heat exchangers in the Next Big Future blog. They claim cooling rates of 1 kW/cm^2.

http://nextbigfuture.com/2006/05/coolin ... imits.html

edit: I'm under the impression that this is being tried with LN2 as a working fluid based on some of the discussions on this site. I am not certain though.

[djolds1]

Suppose H2O is the final cooling mechanism vs LH2? Suppose under such a design you only get 100 MW/m^2. That takes Polywell from 100 MWth to 10 GWth with room for growth. And if LH2 is used as reaction mass for a rocket then we have a viable SSTO vehicle. Especially if a maglev track is used for initial boost.

You propose a two order of magnitude increase in power output without a necessary increase in polywell radius?

IIRC, cited figures are:

100MWth, 3m radius, and
8000MWth, 5m radius

You think we can get 10,000MWth @ 3m radius?

[MSimon]

Suppose H2O is the final cooling mechanism vs LH2? Suppose under such a design you only get 100 MW/m^2. That takes Polywell from 100 MWth to 10 GWth with room for growth. And if LH2 is used as reaction mass for a rocket then we have a viable SSTO vehicle. Especially if a maglev track is used for initial boost.

You propose a two order of magnitude increase in power output without a necessary increase in polywell radius?

IIRC, cited figures are:

100MWth, 3m radius, and
8000MWth, 5m radius

You think we can get 10,000MWth @ 3m radius?

Heat load goes up as the 5th power of radius (if you scale the B field with the radius) so even a larger than 100MWth reactor will have a greater heat load than 1 MW/m^2. I did some calculations on this blog showing why, if the heat load was limited to 1 MW/m^2 then power could only go up as the square of the radius above 100MWth. Very bad.

If you managed to get the power density up the heat load gets worse.

[TallDave]

And yes, it will be probably complex and expensive - but as well be the cheapest part of working fusion reactor!

If you were talking about ITER you would probably be right. But IEC/Polywell needs to be about an order of magnitude cheaper to be economically viable, preferably closer to two.

I am very roughly guessing as described, this would actually be around ten times of the cost of a Polywell reactor, maybe more. Skylon is a $10 billion project, with a $10M - $40M per launch cost. We're hoping 2nd-gen BFRs come in at around $50M.

That said, there could be some useful principles and materials science here that would be applicable.

[TallDave]

I did some calculations on this blog showing why, if the heat load was limited to 1 MW/m^2 then power could only go up as the square of the radius above 100MWth. Very bad.

Especially since costs tend to go up as r^3.

If you managed to get the power density up the heat load gets worse.

Yeah, I think if 100MW BFRs work, huge amounts of resources are going to be thrown at the heat load problem because there's so much bang for the buck there.

[TallDave]

Oh hell. Here's the catch:

The experimental heat exchanger has achieved heat exchange of almost 1 GW/m³,

Read that carefully.

Did you catch that? I missed it first time through. m^3 not m^2. They're talking about cooling a volume of air, not a surface area.

That means

Suppose under such a design you only get 100 MW/m^2

is the best case secnario.

Sadly, I'm guessing this doesn't have much application to cooling a surface.

The microchannels sound promising, though.

[jmc]

Reactions Engines are situated 100 metres from where I work.

[MSimon]

Reactions Engines are situated 100 metres from where I work.

Why not ask them what kind of area heat flow they get per deg delta T?

Part of their high heat flow is due to large delta T. ~1150 deg K.

Commercial work is normally around 200 deg K delta T. or less.

Anything above 5 KW/(m^2 * deg. K) would be very interesting.

This page:

http://www.spiraxsarco.com/resources/st ... onship.asp

assumes 2.5 KW/(m^2 * deg. K)

[MSimon]

That heat exchanger is a miracle of design. However, it is only pumping about 150W/(m^2*deg. K). (I assumed 400 MW and 1000 deg delta T)

Let us go with the figures in this paper:

http://www.reactionengines.co.uk/downlo ... 99-209.pdf

2 GW / m^3. 3000 m^2/ m^3. 500 deg K delta T. Still only about 1.35 KW/(m^2 * deg K)

I can do a lot better than that with water.

Maybe we can do something with internal fins. The idea of course is to get more surface area in contact with the coolant - at least on one side. If we could get microchannel boiling without stagnation or bulk boiling that would be even better.

Once there is a need efforts will be made. I can see getting to 2 MW/m^2 (Space shuttle ME territory) with just a little effort. After that it will get harder.

Some rules of thumb for heat exchangers.

http://www.cheresources.com/uexchangers.shtml

The best case listed is about 4 KW/(m^2 * deg K). With the more typical case being around 1.5 KW.

At 1 MW/m^2 and 200 deg K delta T, I get 5KW/(m^2 * deg K)

1 BTU/(h*ft^2*Deg F) = 5.6782633 W/(m^2*deg K)

The wiki has a number of good external links:

http://en.wikipedia.org/wiki/Heat_exchanger