My Rig in Popular Mechanics

[LTORSINI]

So, I've ogled the pages of Popular mechanics since I was a kid but I really never dreamt I'd build anything myself that would make it to print - but it did in september. Here's the backstory: I've been building liquid cooled computers for about as long as there has been kit to build them, just can't get enough. Back in 2006 I started building Henry - the 1st rig in my profile. Henry started in 2006 as a semi powerful media machine that I used a dodgy Thermaltake Bigwater kit to cool the CPU. He had a quad core AMD Phenom and made some pretty serious heat when I was gaming or... anything really. It made enough heat that I started thinking about how to use it for something other than running up my air conditioning bill.

When I bought and remodeled a house in ~2009 I got a little nuts. I rigged a heat exchanger to preheat the hot water heater and plumbed my man cave with hydronic computer generated floor heat. I also built a heat exchanger and sunk it under the foundation of the house so I could dump any extra heat I didn't need in the summer right into the ground. That really worked like charm but I was only pushing about 700W of electricity through my rig and only recovering about 50% of that through the loop, not really enough to make a dent in the winter heating load. And the thermaltake kit was really not up for the job.

So I went a little nuts and built a monster of a rig out of a ton of ASUS/AMD/EVGA and EKWB kit (sorry Koolance... we tried and failed). It'll pull 1600W fully loaded and makes some serious hot water. Once I got it together and started to put it through it's paces I was really hooked, so I took the last 2 years to optimize this thing to do combined heating and computing. Along the way I melted a lot of watercooling kit and had to reengineer just about everything in the rig to withstand the unbelievable temperatures it hits. I've replaced all the poly tubing with high temp silicon wrapped with coils to keep it from rupturing. I machined new brass covers for all 6 EK water blocks, because they started to warp and weep at about 160f. And after melting the CPVC in my thermal storage system I re engineered the thing out of copper and steel as well as replacing the water I was using with some amazing phase change material (fatty acids and esters - think soap).

Once I had hardened the system so it would withstand extended jaunts above 160f, I had to split the loop and run the north/south bridge, RAM and the two liquid cooled AQUERO 6XTs on their own lower temperature system. Now the primary loop will exit the rig at 200f and we've hit 190F in the thermal storage tanks - something I certainly thought was not possible when we started... and I'm sure many of you are going to tell me it's not. The rig runs so hot that it makes steam across the GPUs. YOu can see a detailed build sheet in my ModsRigs profile and a video of the thing cycling through a heat cycle on FLIR camera

I'm sure there will be plenty of folks to let me know that:

• won't/doesn't work - ie. I'm full of it
  
• is a lame idea - why on earth would you want to run a computer hot?? (there are actually very economic and environmental good reasons - we'll see if anyone wants to talk about them!)
  
• breaks the laws of thermodynamics - it doesn't, start here and I'll help out with specifics
  
• is a copycat of Microsoft, Cloud & Heat, Quarnot Computing or Nerdalize - it's not
  
• Popular Mechanics are idiots - Try New Scientist, Network World, New York Times, OZY or PRI.
  
• Kickstarter was a bad idea - yes, kickstarter was a bad idea. 😉
  
• 
  

[xeromist]

LTORSINI wrote:
why on earth would you want to run a computer hot?? (there are actually very economic and environmental good reasons - we'll see if anyone wants to talk about them!)

OK, I'll bite on this one. I realize spending less on cooling has an immediate economic benefit but deliberately running components hot will shorten their lifespan. If you are lucky the parts will get replaced before wearing out anyway, in which case the difference is moot, but if one of the components was marginal to begin with you might have to replace something at higher expense than an extra radiator. I know yours might be a special case due to the nature of your project but I'm not sure running hot applies to just anyone.

Data centers have shown energy improvements running hot isles or cooling with unchilled water but that's because they reduced hot spots. So while things in general are running hotter, the individual components still aren't overheating.

Also, if the purpose is to capture and use the heat then attempting to store it at high delta T is going to result in increased environmental loss. However, I'm not sure what your material is and it's optimal temperature range so I realize that might limit you. So in your case is it a lack of usefulness at lower temp, sub-optimal phase storage, or limited storage capacity? Just curious how you came to this configuration.

Also, congrats on making print! Any day an enthusiast build makes print is a good day. 🙂

[LTORSINI]

xeromist wrote:
OK, I'll bite on this one. I realize spending less on cooling has an immediate economic benefit but deliberately running components hot will shorten their lifespan. If you are lucky the parts will get replaced before wearing out anyway, in which case the difference is moot, but if one of the components was marginal to begin with you might have to replace something at higher expense than an extra radiator. I know yours might be a special case due to the nature of your project but I'm not sure running hot applies to just anyone.

Data centers have shown energy improvements running hot isles or cooling with unchilled water but that's because they reduced hot spots. So while things in general are running hotter, the individual components still aren't overheating.

Also, if the purpose is to capture and use the heat then attempting to store it at high delta T is going to result in increased environmental loss. However, I'm not sure what your material is and it's optimal temperature range so I realize that might limit you. So in your case is it a lack of usefulness at lower temp, sub-optimal phase storage, or limited storage capacity? Just curious how you came to this configuration.

Also, congrats on making print! Any day an enthusiast build makes print is a good day. 🙂

Great questions.

First, no... I'm not suggesting the average gamer runs their rig this hot. This rig is part of an R&D project we're working on so it is actually designed to run hot.

Yes, running parts hot can/will dramatically shorten their lifespan so you have to be mindful of how you run them and the economics behind the reason to run them hot. The rig you see here has run for about two years at temps up to 190 in the storage tanks and 215 in the loop so there are ways to keep them alive and there are chips that actually run far, far hotter. You are correct that the economic model is to run hard and replace the components on shorter intervals.

Some data centers are running more efficiently these day but the average datacenter still uses between 30% and 60% of this electricity it consumes for cooling. When you consider that 2-3% of all the electricity produced in the US is being consumed by data centers that don't have a need for that heat and that 30% of the total energy we consume in the US is used for heating buildings - you see a match. Eliminate the waste from refrigerating servers and displace the energy spent for one purpose, heating or computation.

The need for heat is pretty seasonal so increasing temperature and running absorptive chillers/refrigeration or regenerating desiccant for evaporative space cooling will extend the system's benefits through the summer months and increase the benefits. It may not make the economic break even, that's why we are doing the research. It is an absolute slam dunk for DHW, much cheaper system. We have a pretty efficient thermal storage system.

[xeromist]

So, how much environmental loss? Is the tank warm to the touch?

[LTORSINI]

xeromist wrote:
So, how much environmental loss? Is the tank warm to the touch?

Current prototype, Henry, only recovers about 70-75%. Take a look at DareHenry, he should recover and store in the 90s.