I'll be more mathematical. By Carnot's theorem, the maximum possible efficiency of a heat engine = 1 - (T_cold/T_hot). Using very rough figures, T_hot = body t

I'll be more mathematical. By Carnot's theorem, the maximum possible efficiency of a heat engine = 1 - (T_cold/T_hot). Using very rough figures, T_hot = body temperature = 37 C = 310 K and T_cold = average temperature of the earth = 16 C = 289 K. Thus, the maximum possible efficiency is 1-289/310 = 6.8%. To get 99% energy recovery, assuming human body temperature remains the same, would require that the cold side of the heat engine be at 3.1K, or only slightly warmer than the background temperature of space. The only possible way to achieve that difference is by radiative heat transfer to space. However, we can use the temperature of the buried lunar regolith as an estimate of what the the "average" temperature of the earth could be without an insulating atmosphere. http://www.asi.org/adb/m/03/05/average-temperatures.html http://www.asi.org/adb/m/03/05/average-temperatures.html says it's -35 C = 238K. Even if we could use -35C as the cold side of our heat engine, that would still only be 23% efficient, assuming maximal thermal efficiency and, you know, no air.