Re: Cooling technique for Jupiter brains

From: Robert J. Bradbury (bradbury@aeiveos.wa.com)
Date: Tue Feb 08 2000 - 08:32:15 MST


On Tue, 8 Feb 2000, Spike Jones wrote:

> > Spike Jones wrote: Each particle would be in the
> > tunnel only about 20 seconds. Assuming they went in 50 kelvin
> > cooler than the planet and heat transfer by radiation alone, would come
> > out only a fraction of a degree warmer than it went in.
>
> Doooh! 200 seconds, not 20! Whats an order of magnitude
> between friends? See what time of the morning I get up to
> correct my own error? {8^D spike
>
"A dedicated scholar be you Spike", says Obie-Wan Bradbury.

I had wrestled a bit with this problem a bit previously and
what you need to deal with are heat capacity and heat transfer.

You need to do something like bring in liquid hydrogen cooled
almost to absolute zero, let it flow though channels in the JB
letting it boil, then put it into containers that are ejected
on the other side of the JB that reliquify the H2 in space
and orbit back around to the LH2 dropoff point. Or, since there
is no atmosphere required for a JB, you could use something like
Moravec's rotating skyhooks to do continual momentum transfer
between orbiting radiators/condensers and the surface. The
interesting thing about these approaches is that they do let
you separate the radiator temperature from the CPU operating
temperature to a large degree.

The only way I've seen to do the heat transfer effectively is using
liquids. Even in the orbiting CPU-satellites of a MB, you still
consume a sizeable fraction of the power for CPU to radiator
coolant circulation. After Anders paper came out, I thought
he might solve the JB problem by having really *huge* coolant
pipe (reducing the friction and therefore power requirements
somewhat), but the real problem is nano-CPU cooling. You have
all of those small capillaries through the nano-CPUs and
you are going to get a lot of friction pumping the phase
change coolant through. The only possible "out" for that
I've imagined might be a superfluid coolant like liquid He.

Another question that comes to mind is *what* are the heat
transfer properties of a Bose-Einstein condensate? Since
I'm under the impression it functions as a "single" atom,
it ought to have fantastic heat conduction values over
a rather large volume. Better than diamond?

Robert



This archive was generated by hypermail 2.1.5 : Fri Nov 01 2002 - 15:26:41 MST