Leeching ambient atmospheric heat: is this feasible?

[DrMrLordX]

One thing I have wondered about is whether or not it would be feasible to use the temperature differential between the atmosphere (preferably in some equatorial climate where it stays warm most of the year, or what have you) and the ground to power thermoelectric generators.

The goal would be to put a massive array of fins of some material or other (aluminum) in the air, and then have another such array around 10m underground, and connect both arrays to contact plates that would be mated near the ground with one or more thermoelectric generators sandwiched inbetween. If done properly (with heatpipes, thermal interface material, and the like), heat should be constantly moving from the air to the upper array of fins, to the upper contact plate, through the thermoelectric generator(s), down to the lower plate, and then into the lower fins and, through them, into the earth. A portion of said heat would be converted to current by the thermocouples (presumably bismuth telluride p and n junctions).

Unfortunately, the delta T between the upper contact plate and the lower contact plate would never be very high. The lower contact plate should remain at around 10-13C all year long so long as the heat load coming from the upper array was never too great, while the atmosphere would vary with weather conditions, night/day, and so forth. The amount of heat the upper fin array could realistically absorb would also be affected by barometric pressure (I would think), since that would affect atmospheric density.

You could probably help things along a bit by sealing the upper fin array under greenhouse glass (or similar material) and by filling the sealed enclosure with carbon dioxide, water vapor, nitrous oxide, or whatever else would help act as a greenhouse gas, though if heat is being leeched out of the air quickly enough, that might not help.

 

[CycloWizard]

Yes, it would definitely be doable, but the amount of energy generated would virtually never recoup the initial capital costs of the project. Using a giant Peltier would probably increase the efficiency by quite a bit, but even on a small scale, the payback period for energy generation using this method is like 3000 years (based on a cost analysis of using a Peltier to power a CPU fan using heat generated by the CPU).

 

[DrMrLordX]

Interesting. If I may ask, what is the best way to model the potential output of a given number of thermocouples?

The specs on theromelectric generators I've seen seem to have a maximum output per # of thermocouples, with actual output based on delta T. Logically the heat source would have to provide enough heat and the "cold source" would have to sink enough heat to move at least enough heat across the thermocouples to account for the power generated.

Then there are layman's science articles that usually describe thermoelectric power generators in terms of their efficiency, which for bismuth telluride p/n junctions is typically said to be 5%.

So the amount of heat moving across the thermocouples might have to be twenty times that of the anticipated power output.

However, depending on the heat source and heat sink, you could be moving the same amount of heat across the thermocouples at different delta Ts . . . which leads me to believe that my understanding of thermoelectric generators is insufficient. I've only played with one once, and I couldn't get any measuable current running through one just putting ice on one side, but I COULD get measurable current running through one by holding it near a running truck engine (old Ford v8). Not that that means much, per se.

 

[CycloWizard]

I would start with this:

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

My textbooks seem to indicate that the voltage V of a thermocouple is usually well described as a cubic function of temperature T,
V=AT+1/2*B*T^2+1/3*C*T^3
with coefficients (A,B,C) depending on the materials used in the thermocouple, as well as the temperature. The voltage is tens of millivolts even when the temperature difference is thousands of degrees, with the highest voltage (~70 mV) occurring in the Chromel-constantan (type E) thermocouple at a temperature of about 1500°F at one junction with the other at 32°F. In other words, you're not going to get much current.

 

[silverpig]

TEGs are horribly inefficient. Peltiers get you 1-2% typically.

If you're trying to get heat from the earth, we already do that - geothermal. But we go far far down where it's actually hot. Part of the reason we do this is to get the high temperature differential. You can't do very much if the temperature difference isn't very high.

Rather than go with a TEG, you'd probably be better off going with a stirling engine.

 

[DrMrLordX]

I would start with this:

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

Yeah, I've read through that. It all makes sense until I start thinking about how the actual heatload never seems to factor into the equation.

The voltage is tens of millivolts even when the temperature difference is thousands of degrees, with the highest voltage (~70 mV) occurring in the Chromel-constantan (type E) thermocouple at a temperature of about 1500°F at one junction with the other at 32°F. In other words, you're not going to get much current.

Apparently uranium makes for a fairly good p-junction . . . too bad it undergoes radiocative decay, even as u235.

TEGs are horribly inefficient. Peltiers get you 1-2% typically.

So I have noticed.

If you're trying to get heat from the earth, we already do that - geothermal. But we go far far down where it's actually hot. Part of the reason we do this is to get the high temperature differential. You can't do very much if the temperature difference isn't very high.

Google has apparently invested in a newer form of geothermal that involves injecting water far underground using deep drilling techniques developed by the petroleum industry. I don't think your average joe can muster drilling efforts that deep, though.

Given my current circumstances, I don't have any property to drill or dig in anyway, so the point is rather moot.

Rather than go with a TEG, you'd probably be better off going with a stirling engine.

Yes, I would imagine so. Upkeep might be higher than on TEGs but output would also be higher.

 

[CycloWizard]

Yeah, I've read through that. It all makes sense until I start thinking about how the actual heatload never seems to factor into the equation.

The voltage generated by a pair of bimetallic junctions at different temperatures is a material property of the two metals. I'm not sure where you were expecting the heat to come into play, but maybe I'm misunderstanding your scheme.