LN2 cooling concept idea

[KidKid]

Would it be possible to get a liquid nitrogen generator and hook it up to a computer (that's well protected with foam and all that to keep it from short circuiting)? If so, that would be a pretty neat thing for LTT to make a video on and possibly beat records with. It would be much easier to cool a PC with LN2 that way... assuming you have the money.

[OhYou_]

the problem with ln2 cooling is it's traditionally done by just dumping it in a pot and the evaporation causes the cooling. If by hooking up a ln2 generator, you mean like pumping ln2 through something similar to a traditional cpu water block, I'm not sure if that would be effective because it isnt evaporating or really its boiling but still.

you'd spend less effort directly cooling the component with phase change cooling. the problem is that most commercial units are smol, and messing around with refrigerants to try to diy it is ???

[Kilrah]

-> Moved to Custom Loop and Exotic Cooling 

 

So your idea is to use an LN2 cooling setup like what exists, but instead of using a storage tank you can actually handle connect it directly to a massive plant?

[GarlicDeliverySystem]

closed loop or semi-closed loop systems have been shown at exhibitions in the past, look for videos on their channel or on DerBauer's channel, GN and Jay also highlighted them I think.

 

The fundamental issue is that LN2 is not that easy to work with in a 'sustainable' way, for a few reasons:

• it is really, really cold. Meaning machinery to produce it is complex, as you have to go through multiple stages to cool air down due to no single refrigerant being able to just cool it. It also means a lot of heat is produced and electrical energy is required.
• This also means that overall cooling power (i.e. how many liters of LN2 you can produce per hour) is quite limited, unless you go to industrial scale again.
• Working with LN2 can be quite dangerous, as the boiling point of liquid oxygen is higher than that of LN2. Meaning if you are not careful, you can create puddles of liquid oxygen in your system. That is not good.
• You create a lot of gas when LN2 is used for cooling, since it is immediately boiled off. So you would need to have a recuperation and compressor system that can handle that amount of gas production. Again, some dangers and complexity involved here. 

[KidKid]

9 minutes ago, Kilrah said:

-> Moved to Custom Loop and Exotic Cooling 

 

So your idea is to use an LN2 cooling setup like what exists, but instead of using a storage tank you can actually handle connect it directly to a massive plant?

 

9 minutes ago, OhYou_ said:

the problem with ln2 cooling is it's traditionally done by just dumping it in a pot and the evaporation causes the cooling. If by hooking up a ln2 generator, you mean like pumping ln2 through something similar to a traditional cpu water block, I'm not sure if that would be effective because it isnt evaporating or really its boiling but still.

you'd spend less effort directly cooling the component with phase change cooling. the problem is that most commercial units are smol, and messing around with refrigerants to try to diy it is ???

 

6 minutes ago, GarlicDeliverySystem said:

closed loop or semi-closed loop systems have been shown at exhibitions in the past, look for videos on their channel or on DerBauer's channel, GN and Jay also highlighted them I think.

 

The fundamental issue is that LN2 is not that easy to work with in a 'sustainable' way, for a few reasons:

• it is really, really cold. Meaning machinery to produce it is complex, as you have to go through multiple stages to cool air down due to no single refrigerant being able to just cool it. It also means a lot of heat is produced and electrical energy is required.
• This also means that overall cooling power (i.e. how many liters of LN2 you can produce per hour) is quite limited, unless you go to industrial scale again.
• Working with LN2 can be quite dangerous, as the boiling point of liquid oxygen is higher than that of LN2. Meaning if you are not careful, you can create puddles of liquid oxygen in your system. That is not good.
• You create a lot of gas when LN2 is used for cooling, since it is immediately boiled off. So you would need to have a recuperation and compressor system that can handle that amount of gas production. Again, some dangers and complexity involved here. 

I was thinking something like this - https://minnuogas.com/nitrogen_plant/liquid-nitrogen-generator/ and hook it up to a PC somehow. I don't exactly know how, but I'm pretty sure @AlexTheGreatish can think of something great and sketchy for it.

[OhYou_]

16 minutes ago, KidKid said:

 

 

I was thinking something like this - https://minnuogas.com/nitrogen_plant/liquid-nitrogen-generator/ and hook it up to a PC somehow. I don't exactly know how, but I'm pretty sure @AlexTheGreatish can think of something great and sketchy for it.

the cost of the big one which might be suitable for continuous use is already $63,000 for just the equiment, and installation costs could easily exceed 100k. I'm not sure if you can get away with less than 5L/h to run a pc, but the investment alone would make 0 sense for something like a youtube channel which will maybe monetize 3 videos about it and it sits dormant. far cheaper to simply buy it by the tank for those oneoff videos.

ths may help
https://minnuogas.com/cost-calculation-case-study-for-5l-hour-liquid-nitrogen-generator/

[GarlicDeliverySystem]

11 minutes ago, KidKid said:

I was thinking something like this - https://minnuogas.com/nitrogen_plant/liquid-nitrogen-generator/ and hook it up to a PC somehow.

That thing in its largest version produces 120L of LN2 per day. Heat of evaporation for LN2 is about 160kJ/L, which is ca. 44Wh. A CPU that runs at 200 would run 12 minutes on that (approximately) , so you would need 4.5L per hour just for a 200W CPU under load.

120 / 4.5 = 26 hours, so that would barely meet the requirements. Issue is you won't have just 200W to cool, you will also have a lot of losses at cooldown and then some more in between. You likely won't run this at 200W under load (after all, what is the point of LN2 cooling if you don't put a RTX5090 under LN2 as well), but you also won't run the thing full tilt 24/7. So an average of 200W seems fair for a 6-8 hours of use a day.

 

The thing you linked pulls 15kW, which is turned into heat and noise. You will need to get rid of that. You will need either lots of bottled, relatively clean N2 or another N2 gas generator. 

 

For what?

[wanderingfool2]

20 minutes ago, KidKid said:

I was thinking something like this - https://minnuogas.com/nitrogen_plant/liquid-nitrogen-generator/ and hook it up to a PC somehow. I don't exactly know how, but I'm pretty sure @AlexTheGreatish can think of something great and sketchy for it.

What @GarlicDeliverySystem said, the LOX is a concern when doing something like this.  Along with the fact that you need proper ventilation to prevent yourself from displacing all the oxygen in the room with nitrogen.

 

Based on 

he was saying he uses 40 - 50 liter for a 4-6 hour period.  You realistically can't generate that much that quickly with anything that would be practical.  The link you posted, that generates at best 120L per day....so if you assume 40 liters (low end) for 6 hours (high end) and 120L (high end) generation per day (all these are what I can figure as "best" case), you get 18 hours runtime per day.

 

But worse case, 50L, 4H, 10L/day.  You get a runtime of 48 minutes.  The 10L/day they say the cost for the machine is $27,000 as an fyi as well

[Hinjima]

Probably not exactly what you are after but was a good concept by EVGA back at Computex 2018

[KidKid]

34 minutes ago, Hinjima said:

Probably not exactly what you are after but was a good concept by EVGA back at Computex 2018

I think that's what I meant. Thanks!

[OhYou_]

1 hour ago, GarlicDeliverySystem said:

That thing in its largest version produces 120L of LN2 per day. Heat of evaporation for LN2 is about 160kJ/L, which is ca. 44Wh. A CPU that runs at 200 would run 12 minutes on that (approximately) , so you would need 4.5L per hour just for a 200W CPU under load.

120 / 4.5 = 26 hours, so that would barely meet the requirements. Issue is you won't have just 200W to cool, you will also have a lot of losses at cooldown and then some more in between. You likely won't run this at 200W under load (after all, what is the point of LN2 cooling if you don't put a RTX5090 under LN2 as well), but you also won't run the thing full tilt 24/7. So an average of 200W seems fair for a 6-8 hours of use a day.

 

The thing you linked pulls 15kW, which is turned into heat and noise. You will need to get rid of that. You will need either lots of bottled, relatively clean N2 or another N2 gas generator. 

 

For what?

keep in mind the working day is at most 10 hours, so it is effectively 10L/h assuming they have a storage tank.

[GarlicDeliverySystem]

4 minutes ago, OhYou_ said:

keep in mind the working day is at most 10 hours, so it is effectively 10L/h assuming they have a storage tank.

Yes, but the 4.5L are just for the power of the CPU. You will have losses along the way, every bit of the machine that is connected to the outside will have small but meaningful losses over time. Put a GPU with 600W on top of that and you have quite a lot of liters per hour during continuous runs like rendering or simulation workload.

 

It could be done somehow, but the economics is really not there, not even for a video. Easier to go with two or three 240L dewars and call it a day. 

 

Other issue that needs to be solved (not sure how they did it in the examples above, I haven't watched the full videos) is how you ensure adequate heat transfer and circulation of the LN2/N2. The expansion from LN2 to N2 yields several hundred liters of gas per liter of LN2, that need to be trapped, stored and compressed for the liquefaction machine (unless it just used a cold head and condensation).

Alternative is to go with two loops at different pressures, a higher pressure 'hot' side where the boiling point is elevated, and a lower pressure cold side where the evaporation takes place and cools the high pressure hot side. However, you would still need to ensure that the liquid in the blocks and hot lines never starts boiling, or you would end up with a lot of pressure and issues for the pump.

[Kilrah]

Eh, basically it'd just be phase change cooling like they did by slashing an air conditioner, just with a way more impractical refrigerant fluid. 

[freeagent]

I agree, I would rather do phase over this.