Topic: Immersion cooling - Beeminer hive - water consumption (Read 319 times)

Beeminer looks perfect for us we are building out a hydroelectric generation and already have river flow to supply perpetual cool water.

How much power do I save on the S19 by removing the cooling fan?

---

Will I still need room cooling with a Beeminer system set up?

---

with Antminer S19j pro
We are budgeting 25% overclock
and 10% reduction in electrical costs from not using fans on the mining rigs

Are those reasonable assumptions?

[moderator's note: consecutive posts merged]

Hi everyone,

I am starting my project to submerge my miners, I have 10, but I am going to test 3 machines first.

My intention is to make the tank with the right measures to make the most of the liquid and cool it with an exchanger with another fluid that will go at a low temperature.

If you want that when I have it done, I will give you more info, tell me

This is thermodynamics 101. I'm running our immersion program and have modeled the CFD of these systems more than once. I'm going to be general here and try to provide more of a road map for those who want to play with their own systems.

Define these parameters:
a) How much heat is being generated on a time basis.
b) Thermal conductivity of selected heat transfer fluid.
c) Specific gravity of dielectric
d) A, B, and C will tell you how fast your flow rate needs to be inside the unit to "refresh" the hot for cold fluid near the chips.

Once all that heat is being transferred to your exhaust system you need to back the math (heat load) into sizing a radiator. You can find these capacities on the units. Dry coolers work great if ambient temps are low enough to allow for sufficient transfer out of the fluid. You can extend your radiator run if you're committed to this route but it gets too damn hot outside to be efficient.


If you're transferring the heat to a water system (to then be transferred to ambient) then presumably you're utilizing a cooling tower system. Your water rate is the evaporation rate. Many, many online calculators for this. The latent heat of vaporization will tell you the remaining energy required to evaporate the heated water, the national weather stations will provide geo-local climate data (and calculators), and the two will define your water loss (refill rate). In many cases the systems provider will know exactly what their capacities and performance expectations are.

Unless you're rocking serious volumes don't bother looking into a condensation recapture system... it's freagin cost prohibitive.

You can consider a variety of hybrid methods. Dry coolers are pretty much always our first line. If summer spikes require greater capacity, we'll end up running a condenser as a second leg to finalize the cooling process after the driers have done all they can. This "subsystem" serves as a redundancy and overflow capacity. To be frank, though, the Texas summer heat can force it to assume a very large proportion of the cooling process during peak.

So. How much water will you lose?
8.33 BTU to raise 1 gal by 1 degree F
Roughly 8100 BTU to evaporate 1 gal. (starting temp)
1 watt = 3.41 btu

(it gets stupidly complicated to define your loss locally by hand. Be general with your RH and velocity variables).


I think an approach to desert miners would be a wet cooler. The heat transfer of vaporization is really about all you can do at those ambient levels before you're back to condensers. Otherwise, bury some geothermal pipe.

You could build a home setup relatively easy, the Beeminer is nothing more than a steel box with a pump and a heat exchanger. It doesn't come with immersion fluid so you'd have to take account of the cost to fill the tub, the dielectric fluid comes in at around $180 per 20 litres.

Phil is right about using an auto radiator as a primary heat exchanger, even a relatively small radiator with 16" push-pull fans can 'exchange' over 50kW to the atmosphere at 35 C, just think for a moment about how many big SUV's there are that operate in hot coutries without a hitch. I've attached a link about a project to make a compact auto radiator that can 'sink' 70kW of heat : https://www.maplesoft.com/support/help/maple/view.aspx?path=applications/RadiatorDesign

It's a bit technical but it shows just how much energy a 18 x 18" radiator with fans can dissipate. It would have to be de-rated a bit to cope with the dielectric fluid though.

I'm sure there are plenty of guys on the forum that could help you build an immersion setup, I'm confident you could make a diy setup to host 20 S19 pros for less than $10k.

Anyone up for a project?

The system I had mentioned getting a quote on used an evaporative cooling tower.  I believe it was more for massive datacenters and not small time usage though, hence the six figure cost.  I'm glad to see that development is being done in this area.  I know I would be willing to spend the amount listed as the sales price without batting an eye to have such a setup.  I wonder though how much cooling would cost that could be implemented silently in a residential environment.  Still, seeing the tank for sale at a reasonable price is a huge step in the right direction in my book.

Any estimate on a total cost with cooling solution?

This looks better than the beeminer as you simply run coolant tank to a radiator and do not need a water based system to cool the coolant system

Updated post: Im a new member to Bitcoin Forum,

Here's an article to read: https://www.engineeredfluids.com/post/corrosive-sulfur-in-mineral-oil-a-killer-app-for-electronics

Its an interesting subject on cooling but its not without questions, one issue on corrosion may lead to increase  operational costs.

Then theres' this article:  https://www.engineeredfluids.com/for-crypto-currency

Video: https://www.youtube.com/watch?v=PvmMs6mU0N

Its important to note in the video, their using Di electric coolant Vs. mineral oil

This video is for the purest whom will prefer to see ASIC working in coolant: https://www.youtube.com/watch?v=441JEdMKVSw

very interesting piece of equipment.

https://beeminergroup.com/en/catalog/beeminer-hive-12rx/



They mention a closed water circuit can be used. But you are in a desert so using a pond is not an option.

Neither would using a swimming pool as I think the water in the pool would get too hot.

Now if you dump the coolant heat into a closed water system with lots of truck radiators and fans maybe it works.


https://www.ebay.com/itm/403085000869?


so beeminer with immersion liquid exchanges the heat into multiple truck radiators. the truck radiators need shade from sun and tons of fans

you need to stay under 50c. and what is the desert max temp?  (120°F − 32) × 5/9 = 48.889°C
So you could have an issue due to the suns heat. I also know  my estimate of max temp could be off by 2 or 3 °C

I mean Beeminer Hive uses cold water that comes to the heat exchanger - https://beeminergroup.com/en/catalog/beeminer-hive-12rx/

In our location running cold water is very expensive. How did you do your setup? Especially oil cooling part?

Immersion cooling is like the holy grail of mining.  Very few pull it off successfully, but we all want to do it.  I've had installation quotes in the low six figures to get a finished product installed in my garage.  It isn't a simple thing to pull off.  Maybe you mean liquid cooling, which is like using a radiator to cool, where immersion cooling is submersing your miner in liquid?

In any event, if you do make an immersive mining model that uses "water" to keep it at cool temps, be sure to have a video camera recording when you flip the on switch.  I'd like to see the ensuing fireworks show.

Hello miners!

We are working on the model for a mining farm and would like to learn from you regarding immersion cooling for Antminer S19 Pro 110 THs/ and/or S19j Pro 100 TH/s, with Beeminer Hive setup.

Basically, our question is: "X amount of water to keep it at Y temp"

Maybe you have some experience? At least in our model, the water cost looks super high...

Any feedback is appreciated. Many thanks!