Harmonics don't really effect ASIC and solid state systems. For example, the only concern you need to have with harmonics and vibration is with spinning disks (magnetic disks) in a traditional DC environment. Bitcoin miners have none of these problems. As for the rest of it, I won't even bother addressing as Helmholtz resonance is completely irrelevant to anything really... No part of the HVAC system has empty cavities, the hot isle exhaust most of the air out of the facility except for what we wish to recirculate over winter..
There is also plenty of space in the hot and cold isles to handle the volume of air that is passing. We're not using high pressure 36'' (3ft by 3ft) ducts, these are 6ft by 14ft ducts...
The design of the PSU to be above the miner is actually imperative, as bitmain supplies such short DC supply cords. The entire weight of the PSU is supported by two rails that the PSU slides into, and the PSU itself is simply screwed into the front to allow someone to unplug / re-plug the power cord if they need to simply power cycle a miner without having to go into the hot isle...
"The corrugated cardboard + plastic foam + particle board can direct the air around metal warehouse" - yeah and cardboard and plastic have one thing in common, they are both combustible. Regardless if I were stupid enough to do it this way and risk burning thousands of dollars worth of equipment to a crisp, the firecode does not allow me to, nor does the facility itself, so both 'cheap' solutions are irrelevant. If you want to look at what happens when people do this, lookup spoondoolies-tech fire hong kong.
As for multi crew... you obviously are not a very organized person... All S9's we know the hashrate, operation, and the location of every miner in the facility... We keep a database online (intranet) that shows when an S9 goes down, or is hashing slower. A single user can walk to the floor, find the miner, and take it back to the NOC for servicing, often times about a 10 minute operation to swap bad parts and RMA.
As for deployment, yeah, we bring in a team of temps that we pay about 10 dollars an hour to 'install' cable' and do basic stuff. The core team supervises, does final install, and logs the position of the equipment with a tablet on wifi (temporary wifi) pressing the broadcast button one rack at a time. Takes a little while, but once mapped, 1 person working 20 hours a week @ the facility can service 2500 S9 units easily. Its called being organized...
To give you some numbers, the total cost of the solution, compared to the cost of the miners / equipment purchased was 2%. Meaning, 2% of our total cost that we spent on the entire deployment FOR JUST THE MINERS, was the cost of the shelves... The shipping cost was approximately 8% of the total order...
Estimated lifetime for 16nm equipment I give 18 months, before 10nm readily available. With our cost of power 99% lower than the rest of the US, if we're not profitable, then no miners are going to be. at 2.5c per KW, there are only a few places in the world that are cheaper, and nothing else in the US. Technically our old spoondoolies tech miners are still profitable on our current rates, but it would be a waste of space for us to keep them.
Final thought, these things take 5 minutes to install, work flawlessly, and will accommodate future generations of bit-main miners. They work with the S1,S3,S5,S7, and S9... My guess is, they'll work with the S11 as well... How much time labor and money do you spend trying to get the combustible cardboard and plastic up? I've done it before in a cheap facility and its a lot more work than you think.
Obviously you'll go with the combustible datacenter in a chicken coop option, but some of us prefer making money hand over fist and being cheap doing it and not have to worry about 30% of our equipment failing in the first year... I feel many other people are in the same position we are when dealing with multi-million dollar deployments of bitcoin miners. For about a total cost of ownership of 2% greater than fire...
You are 100% correct saying that I've never managed a coin mining operation. I have completely different background. In particular I'm not a mechanical engineer, but I have worked with them. I also don't normally do "operations," but in my company there's a team that specializes in acquiring bankrupt data processing operations, so I've seen quite a few and helped restart them under new management.
I didn't made myself clear in my previous posts. I don't consider your design over-engineering, it is something akin to mis-engineering or show-off-engineering. It is for designed for "lookists" and quick sales, not for the cost-effective operations.
From the mechanical point of view, you've only partially addressed the air flow and the MTTR issue. Your design seems to put the whole weight of the power supplies on the edge of the sheet metal. The better design would be to place the weight of the inside devices on a sort of ledge that is at least 1cm wide and lined with soft rubber. (Sorry, I'm not an ME, I lack proper vocabulary.) Basically your design seriously ignores the vibration and resonances caused by the high speed air flow. The resonances are particularly important when nearly the whole facility will be filled out with exactly identical equipment. Your design also wastes a lot of sheet metal for the bottom of the case (and maybe sides).
The Bitmain miners are already quite similar to the Helmholtz resonator (
https://en.wikipedia.org/wiki/Helmholtz_resonance). The multi-mega-watts facility will be like a mad-scientist punk-rock one-note-organ or maybe one-note-straight-flute ensamble. People already are complaining that they are loud, the vibration of the whole facility will badly affect it mechanically.
The other MTTR issue is with the facility shown in the photo. The very long aisles, over 20 racks long, may look impressive on the photo. They are very detrimental to proper operations: it requires multi-person crews for maintenance. If people working alone are forced to work in it they will either skip important work steps or will spend more time walking around than productively working.
I've seen what an experienced water-damage contractor could improvise in hours to direct dry-air flow around the equipment to baffle and cool it. The corrugated cardboard + plastic foam + particle board can direct the air around metal warehouse shelving no worse than dedicated metal racks and cases. It doesn't look good, but operates well, is much cheaper and has no lead time for ordering. I wouldn't advocate that for permanent facility, but what is going to be the expected operational lifetime of the particular models of coin miners?
