Chris Phoenix deals with waste heat, cooling, and other practical concerns in "Design of a Primitive Nanofactory", at http://www.jetpress.org/volume13/Nanofactory.htm

It's a fundamental limit on manufacturing speed, and isn't going to be solved any more than the speed of light is going to be solved. We're not close to fundamental limits yet. Molecular manufacturing should get us close.

If you look in the Nanosystems index under energy dissipation, you will see much more than one mention, but Freitas provides better summaries in his
Ecophagy paper (discussing how it limits grey goo) and in section 6.5.6 of Nanomedicine.

The problem is well known. In part it is dealt with in Nanosystems Section 11.5 "Convective cooling systems". It is interesting that Keith Henson and Eric Drexler had a patent on the fractal cooling system (U.S. Pat. #4,769,404 – "Heterodensity heat transfer apparatus and method" which is what is discussed in this section.

The problem is further discussed by Robert Freitas in Nanomedicine Volume I Sec. 6.5.7 "Global Hypsithermal Limit" (which deals with the limits of heat dissipation from a planet such as Earth. This sets a maximum upper level on the amount of nanomachinery you can have operating on a planet — it works out to about 10 kg/person. The only way offset this build *very* large radiators to radiate the heat into space. I've discussed this somewhat in my planetary dismantlement paper.

The short form is that diamondoid is a very good heat conductor. The heat moves through the diamondoid into a circulating fluid (think automobile or nuclear reactor cooling systems) which then moves the heat to a radiator or cooling tower of some sort. There is nothing new to nanotechnology in this respect. What nanotechnology adds is things like fractal plumbing, perhaps higher velocities in fluid transport, phase change coolants, etc. But it is known that there are still limits. One might compensate for some of the limits by developing more efficient chemical reactions or processes for whatever tasks are being performed so less heat is produced. For example — use reversible computing. It is an area of active research and produces much less heat than current computing methods.

Specification of a complete cooling system requires a relatively complete specification of the nanomachines themselves as well as extensive computational capacity to model the fluid dynamics simulations that need to be done. These were beyond the scope of Nanosystems. But there can be no doubt that at least for some nanomachine designs there will be a feedback component in their design between the machinery that performs the desired function and the capacity of the cooling system. This work would only be different to a small degree from studies of how you keep an automobile engine or the reactor in a nuclear submarine from overheating.

Robert