Quote from: 2112 on June 26, 2014, 03:06:00 PM

As usual antirack is omitting the cost of:

- water tower
- water pump
- water circulation pipes
- fans in the water tower

Quote from: 2112 on June 26, 2014, 03:06:00 PM

omitting important necessary items in the cost projection.

Quote from: 2112 on June 26, 2014, 03:06:00 PM

Again, my argument is not only about cost of the Novec fluid. It is about a sum of costs of Novec and the whole secondary loop.

Quote from: 2112 on June 26, 2014, 03:06:00 PM

Actually antirack was intentionally omitting the costs and dimensions of the secondary loop.

I do not understand why you believe that I am omitting or hiding anything. Or why you believe that Allied Control or DataTank Mining does.

These parts are included the price stated ($0.50/W), included in the efficiencies stated, etc.

The system works as described and the cost is as described, with all parts included. It's a turnkey solution.

That is the value proposition of the system. It costs less than other solutions. Why is this so difficult to understand? Why do you need to look for hidden costs where there are none?

The cost is included. For _everything_, including the Novec fluid, the secondary loop, and everything else stated on the site, in the prospectus, etc. it also includes the site preparation (ie. power feed in place, transformers, etc). DataTank Mining prospectus: "full infrastructure cost, including site prep". There is nothing hidden. Also as stated in the prospectus.

I can understand why this all might sound confusing if you are not involved in building such infrastructure or running a facility, after this is all industrial stuff not buying and connecting some ATX power supplies from Newegg.

But I am not sure why you would think I am hiding that "fact" - rest assured it's not my intention.

About the Novec, the work around today is to build dense hardware, and I believe we are just on the right track ("1,000kW of mining hardware in a volume smaller than the bed of a pickup truck"). 3M is not the only company selling Post-It's, why would they remain the only company selling fluids? At this moment 3M seems to be the best, for various reasons. DataTank works with any other fluid too, they don't mind which magic juice you put in them.

Back to the coolers you believe I am omitting:

You can see the "machines" right here, so there is no questions that they will are required.

If you read the paper "Bitcoin, 2-Phase Immersion Cooling and the Implications for HPC" that was also published in Electronics Cooling, you can basically find a user guide how to build your own system.

I write "the custom tank condensers use enhanced tubes borrowed from the refrigeration industry. Each is fed facility water directly from a four module dry tower sized for Hong Kongs hot and humid weather (Table 3)." And there is data.

http://www.allied-control.com/publications/Full_Version_Bitcoin_2-Phase_Immersion_Cooling_and_the_Implications_for_HPC.pdf

Here are some other links to show how it works, and of course a radiator is required.

http://www.allied-control.com/blog/immersion-2-tech-details-cooling-hundreds-of-kilowatt-with-1500-watt

(you'll see right there in above's blog post that the closed loop water system is a vital part, and how to design it, and again pictures of dry coolers too)

Excerpt:

Quote

Oversized Closed Loop Water Circuit

In order to accommodate a possible increase in capacity, we have built a simple closed loop water circuit. When sizing the flow, pipe size and other properties of the circuit, our engineers intentionally went a couple of sizes larger than what is technically required. Water volume is around 1200 liters and it's a closed loop water system without any chemicals, no water is being wasted.

The extra cost of using larger pipes is minimal, especially in view of the gained extra lifespan and room to for much higher IT loads. As a side effect, the efficiency of the water system is greatly improved, as larger systems run smoother, which leads to further energy savings.

Silent Operation

Our system automatically slows the dry cooler to the minimum required fan speed, which is at this time of the year (around 25°C outside air temps) just 10-15Hz. The result is a dry cooler which operates below the background noise levels. Besides, the dry cooler employs fans that borrow their blade design from birds and aero-acoustic research. The same can be said for our water pumps. With only one of them online at a time, they rarely run more than 25Hz and the noise level when standing right next to them is less than that of a small 3HP outdoor air condition unit.

If you build your system at a source near free cooling (ie. a cold river), you could replace the dry cooler with a simple "garden hose" that runs up and down a few stretches in the river. This again is mentioned in the prospectus:

Quote

availability of free cooling may reduce cost by automatically disabling cooling equipment, eliminate cooling modules, or allow sharing of cooling module with several clusters.

If you replace your 'dry tower' with a wet tower, you could make it even more efficient, because now you have evaporation a second time, but now you are evaporating a lot of water (Facebook uses liters of water per day per kW!) and how cares at this point if you save (arbitrary, to make a point) 99.75 or 99.43% of energy? And wet towers are a mess, you have to clean them a lot and hence you'll need two (redundancy) or switch off your mine for a half day once a moth.

If you click on the TECH link on the top of http://www.datatank-mining.com you will end up at most of these pictures and documents I am stating here.

https://drive.google.com/a/allied-control.com/folderview?id=0ByWHHc0u_thNMWtQeDNiT2duU0E&usp=drivesdk#grid

Physics are physics.

I am not hiding anything.

Why would I want to mislead anyone about these facts and why you are so hostile Huh

Have you been working on a business plan that was interrupted by DTM (ie. the idea about the large scale water cooling system/cold plate with the smaller PCBs)?

If so, keep physics in mind:

http://books.google.com.hk/books?id=0t-HrUf1aHEC&pg=PA20&lpg=PA20&dq=100,000+W/m2-K.&source=bl&ots=tCOzLR5cVJ&sig=v-nA8-lRSBwZ4ZOAqwi3rGVdhsE&hl=en&sa=X&ei=Pf8hUe2iNfCiiAePlYH4AQ&redir_esc=y#v=onepage&q&f=false

Translation: 2-phase is more efficient. 2-phase is passive (= zero energy spent). Water cooling is less efficient. You need to spend a lot of energy to make it happen. That's why 2-phase cooling is employed, not only in DataTanks. For decades. But open bath is new. And Bitcoin densities Wink

There is no reason for bashing each others solution, they all have their place and even air cooling won't go away anytime soon IMHO. Go ahead build yours, only after we all build our large scale implementations we can tell which one will actually do a better job.

Either way, since you replied to the post about the Spondoolies Tech SP30, I am saying they should dump it all in liquid immersion cooling tanks and use passive 2-phase immersion cooling, and you are saying they should build smaller PCBs and mount it all on a giant heat sink. Both obviously saves them a lot of money.

I'd like to request to take this to a separate thread so we don't spam the DataTank Mining thread with technicalities. There are threads about 2-phase immersion cooling.