12 Feet Under: 1000 Square Feet of Geothermal PC Cooling
After a couple of years of wanting to do a deep geothermal cooling loop everything has come together to make it a reality. I kept convincing myself I I could settle with a few giant radiators and a huge case but this project's components are actually quite a bit cheaper. The 300 feet of geothermal pipe coil and the Iwaki pump together were less than a 560mm SR-1 radiator for example.
All the major components are already on the way or already here, and the excavator is nearby and ready to go. However, the continued wet weather will most likely delay digging the trench until it stops raining or snowing twice a week. This will make a big enough mess in my yard and the hill my house is on as is.
I also need to order the 1-wire temperature components and wired pulse flow water meter to gauge the effectiveness of this project. Some extra time to get that all planned, wired, and programmed before digging will be helpful.
Logging and testing the cooling capacity year round will be where most of the fun is at in the end. For example, I'm interested in seeing if the loop temperature changes while going down to and from the trench coils. Black PE pipe insulation and strips of blue foam will be used on all the exposed and shallower buried piping to keep as much of the outside ambient temperature changes from affecting the loop. The whole point is to capture the deep grounds relatively steady temperature year round. Letting it get too cool in the winter, or too warm in the summer defeats that.
The current loop layout is described in the following image. The ground loop coils aren't shown at the bottom but the text explains them. Google geothermal coil loops if you don't know what they look like installed in a trench. They're a bit hard to draw pixel by pixel.
http://www.bibliognome.com/pictures/12feetunder.gif
I'm conveniently located in Virginia whose seasonal ground temperatures are depicted in this image.
http://www.bibliognome.com/pictures/...g-vs-depth.gif
My big unknown is how my much my sites location on top of a deep natural well affect the temperatures. If I delay the digging I can jam pipes deep down with temperature sensors but in the end I'm limited in site location. The drain field, septic tank, and utility wires occupy three sides of the house. The last side is a 15 foot hill. Digging at the top should result in well drained soil, but would cause erosion.
Eventually I might separate the geothermal loop from the PC(s) and instead cool them through copper coils in another tube reservoir. It'll depend on how much restriction I have. For now the simplest way to start is to have one PC directly in the loop as pictured and measure the flow with my king instrument flow meter.
I'll then add more blocks to the loop to see how much flow would be lost from hooking up additional computers in-line. Going parallel with ball valves to control the amounts of flow to each PC is also an option. Each block in the main PC could also be hooked up in a parallel fashion with quick disconnects as well. That way a restrictive (chipset) block wouldn't kill the flow for the others.
The long tube res also makes it simple to dump more heat into the loop through an inexpensive high wattage aquarium heater. If the cooling capacity is as good as I think it will be, I'll need to do more than one to saturate the loops cooling capacity. Live 1-wire logging of this will be GREAT.
300 feet of 3/4" ID geothermal HDPE SDR11 pipe. 3/4" is the sweet spot for price, head loss, and flexibility. Moving to 1" ID wouldn't increase this pumps flow rate much and the increased wall thickness nearly negates the extra cooling surface. This coil has an OD of 1.05 inches or a total surface area of 990 square feet. The amount of soil contact and its temperature is much more important than the thermal conductivity of the material.
Copper has ten times the thermal conductivity of HDPE but its still limited by the soils capacity to remove the heat. 3/4" ID type L copper also costs 10x more, before freight! Going to down to an exact 1/2" ID from this pipes .86 would also increase the head pressure loss from 16 to 160 feet of head pressure. That all made HDPE the only effective choice and cemented the go big or go home aspect of this project.
http://www.bibliognome.com/pictures/phpwclGOf.jpg
To overcome the tubing's friction loss a suitable pumped designed for high pressure had to be found. Luckily surplus Iwaki MD-30RZT pumps available for less than $125 shipped. They have a perfect PQ curve for this project and aren't too thirsty with power requirements at 1 amp max. Look at all of that restriction busting head pressure! I've calculated that half of it will be used to overcome my HDPE pipe's friction and vertical rise. That still leaves 18 feet of theoretical head pressure at 4gpm for the water blocks. That's twelve times more head pressure than a DDC 3.2 at 4gpm.
http://www.bibliognome.com/pictures/30rzt.gif
The 3" PVC tube reservoir (some hack sawing and PVC cement required).
http://www.bibliognome.com/pictures/tubereservoir.jpg
The pump should arrive Friday and the piping should have already been delivered. Unless FedEx lost it in Hagerstown, I'll try to test a simulated loop this weekend. The pump will go upstairs and the coiled pipe will sit outside at the ground level to simulate the trench depth. The flow meter results will be enlightening. I have spare quick disconnects sitting around so I can easily throw in a few blocks or even hook the loop up to my PC and test the indoor ambient cooling while waiting for improved weather. Stay tuned, that's future update material for sure.
