Air/Water Cooling and Efficiency

[imix500]

Hello all, so this is way off of audio but I figured what better place to ask some questions about thermal dynamics and the like. I built an air conditioner a couple years ago and now with the NY summer approaching I'd like to do some work on it.

As it stands now the air conditioner uses about 1500W worth of peltier devices to pump heat away from two large wind tunnel heatsinks I took out of large dc motor drives. Variable speed high cfm fans pull air through them into the room. The heat from the peltiers is pumped into two water blocks, or water cooled heatsinks, and the heat is exchanged into the outside air via a surplus hydraulic oil cooler mounted in the window.

My question is:
If I change the wind tunnel heatsinks into additional water blocks to cool circulating water and use a radiator to cool the air through it, would this be a more efficient approach? It would add a second closed loop water system but I have most of the parts to do it. I'm leaning on trying it, but I wanted to ask for opinions/advice first. Thanks!

[Ron E]

In evaporation cooling technologies, efficiency is basically increased when you bring the cold and hot plate temperatures closer together.

I know very little about peltier devices, but the best way to find these things out is to do a simulation. On a first look, it seems that reducing the hot plate temp would gain you more than increasing the cold plate temp, as an added extended surface would do....

Add a fan to the hot plate radiator or spray water on it....

Measure the hot and cold plate temps and get back to us...preferably with a performance characteristic for your peltier devices.

[imix500]

I should have specified, the hot side of the system works great and I am looking into a few improvements to further it's effectiveness.

My question was more on the cold side and the most efficient way to pull heat out of the air, either via the cooled aluminum wind tunnel heratsink or a radiator and water circulating system.
However, you bring an excellent point. Lowering the hot side temp would proportionaly lower the cold side. My peltiers are around 80W each and there are 16 of them. I don't have exact specs on the modules I have, but the specs for the current models closest to mine are here:

http://www.tecooling.com/ST-127-14-85.htm

[TwoSpoons]

If you want efficiency, throw away the peltiers and get an ordinary pump based air-con. Peltiers have appalling COP. I did a quick calc off the graphs on that webpage, and for a delta T of 10 degrees (reasonable I think) your peltier eats 88W to pump just 64W ! A standard gas heat pump would be transferring more like 190W for 88W of input power.

There's a reason you kitchen fridge does not run on peltiers ...

[AndrewT]

Hi,
evaporative cooling got a mention.

Don't. Legionaires!!!!!

A conventional heat pump can usually achieve between 3 times and 5 times power ratio. i.e 100W of electricity consumed running the electric motor gives about 300W to 500W of cooled air.

Can you find data on a gas powered refrigeration cycle. The cost of the gas is usually much less than the electricity during peak times. Downside would be extra bulk compared to electric driven.

If Peltiers are as bad as 1 to 1 then throw them away.

Old refrigerators may be a cheap source of components.
Would freezers be better, maybe not if they are optimised to produce very cold temperatures at the expense of extra electricity.

[imix500]

Thanks guys,
Well thats unfortunate. That would explain why it's a battle to get anywhere near a differential of even 10 degrees.
So, this is a really basic question I'm sure but I can't recall the answer. So if a gas a/c unit needs 100W of input power, how can it produce 200-300W of cooling power? Where does the extra 200W come from? Thanks.

[AndrewT]

Hi,
you're stealing it from your cold sink (the outside air).

But to allow the steal and sell you need a heat pump which needs a little power to run the pump.

[imix500]

Ah, yes. That makes sense. Well, looks like it's off to ebay with all this stuff. Thanks!