Nov 11, 2019
5
0
10
0
Hi. i am thinking about passive water cooling for CPU+GPU ( i7-9700k + RTX 2080 ti ) estimate overclocked heat load around 600 watt.
I thought about getting a big (40 liter) water tank.
The idea is, that the large amount of water will be able to absorb the heat from my CPU and GPU and the large tank would be able to dissipate the heat from the water.
How much water do I need if I want under 50c for both my overclock CPU and GPU ?
ANYONE Tried It? How big water tank do i need ?
Is possible it works ?
if anyone can make , test please..
the positive side is silent PC and cheaper custom loop
 

Eximo

Titan
Herald
Cheaper how exactly? No radiators?

Seems like a waste of time to me, you could certainly do some calculations to figure out the heat capacity of the water and how quickly it would take 600W to heat it up. The whole idea of radiators is surface area and exchanging heat with the ambient air. Here you would be rely on pure mass, some evaporation, and whatever thermal conductivity the tank has with its environment.

Back of napkin calculation says it will take approximately 4 hours for 600W to heat 40L of water from 25C to 80C. That is assuming a pretty insulated system. Not sure how to calculate passive dissipation of the water tank.

I've no doubt it will work for a little while. Though you would want to keep the water free of debris and it would be need to be bleached or something to prevent bacteria growth. Would also make for a very humid room I imagine.
 

TJ Hooker

Glorious
Herald
You'd still need a pump to circulate the water between the tank and the components.

A sealed tank of water by itself isn't going to do a great job of dissipating heat. And I'm pretty sure it'd need to be sealed so you don't get a bunch of crap ending up in your lines/waterblocks.

How much money do you think you'd be saving with this? You'd still have to buy the water blocks, pump, tubing, fittings...
 

USAFRet

Titan
Moderator
Mar 16, 2013
115,450
2,271
145,090
18,762
You still need a pump to move the "water" past the CPU face and circulate through the tank
How will you cool the water in the tank? Evaporation on the surface? Eventually, you'll just have a tank of hot water. It wil take longer, but it will happen.
You still need fans to move air past all the other components.
50C is zero performance difference than 70C.
 
Reactions: TJ Hooker
Nov 11, 2019
5
0
10
0
Cheaper how exactly? No radiators?

Seems like a waste of time to me, you could certainly do some calculations to figure out the heat capacity of the water and how quickly it would take 600W to heat it up. The whole idea of radiators is surface area and exchanging heat with the ambient air. Here you would be rely on pure mass, some evaporation, and whatever thermal conductivity the tank has with its environment.

Back of napkin calculation says it will take approximately 4 hours for 600W to heat 40L of water from 25C to 80C. That is assuming a pretty insulated system. Not sure how to calculate passive dissipation of the water tank.

I've no doubt it will work for a little while. Though you would want to keep the water free of debris and it would be need to be bleached or something to prevent bacteria growth. Would also make for a very humid room I imagine.
THANK YOU FOR REPLY
how did you calculate the time the water heats to 80 C ?
the idea was to use a closed lid water tank... so no evaporation and no bacteria's..
the question was "how big water tank do i need to keep it cool?" i thought the rooms 25 c will cool down the tank is the tank..
 

TJ Hooker

Glorious
Herald
THANK YOU FOR REPLY
how did you calculate the time the water heats to 80 C ?
the idea was to use a closed lid water tank... so no evaporation and no bacteria's..
the question was "how big water tank do i need to keep it cool?" i thought the rooms 25 c will cool down the tank is the tank..
Heat capacity of water is 4184 Joules per degree C per kg.

[40 * 4184 * (80-25)]/600W = 15341s = 4.26 hours

That's a very rough estimate, as it assumes that no heat in the tank is being dissipated into the room. But the point is that a closed tank of water is not an efficient heat exchanger, meaning that eventually it's going to heat up at which point it won't be able to cool your system anymore.
 
Reactions: Zsoltluu

USAFRet

Titan
Moderator
Mar 16, 2013
115,450
2,271
145,090
18,762
THANK YOU FOR REPLY
how did you calculate the time the water heats to 80 C ?
the idea was to use a closed lid water tank... so no evaporation and no bacteria's..
the question was "how big water tank do i need to keep it cool?" i thought the rooms 25 c will cool down the tank is the tank..
The glass or plastic tank surface is a poor conductor of heat.
 

rubix_1011

Contributing Writer
Moderator
Doubtful. Here's why.

What you are suggesting is a large watercooling reservoir. That's all. The tank would need to be metal or a substance that has the ability to dissipate thermal energy. Ambient room temperature will always need to be equal to or lower than the water inside the tank. It will also need to be protected from growth using some form of inhibitor or biocide. This simply cannot be 'just water'. It won't take long for the warmth of the cooling loop to provide a place for microbials to bloom. The tank also cannot be made of anything that is going to contribute to the introduction of galvanic corrosion.

Second, you need a pump and water blocks to be effective. A good CPU block is between $50-$80 (US dollars) while a good GPU block (full cover) is $100-$140 US dollars. A good pump like a DDC or D5 is going to run you another $65-$85 US dollars. You will need fittings, and even the most basic barb fittings are about $2.50 a piece...you will need at least 2 for every component. You will also need tubing - some of the cheapest is around $1.50 per linear foot.

This doesn't even touch the 'large reservoir' tank you are describing. Where will you source this? What size/capacity? This is very important, and here's why: it MUST be able to dissipate more thermal load in watts than the cooling loop components (CPU + GPU) are producing each second, otherwise, the temperature of the water will continue to rise higher and higher unless load is diminished.

I myself have a 9700k and an RTX 2080 and a fully watercooled loop for them. I can tell you that under load, those will put out some watts.

So, I would say that this idea is not very feasible nor is it done at a low cost, unless you can get much of the above for very, very, very low cost or free.

Do not steal.
 
Reactions: TJ Hooker
Nov 11, 2019
5
0
10
0
You'd still need a pump to circulate the water between the tank and the components.

A sealed tank of water by itself isn't going to do a great job of dissipating heat. And I'm pretty sure it'd need to be sealed so you don't get a bunch of crap ending up in your lines/waterblocks.

How much money do you think you'd be saving with this? You'd still have to buy the water blocks, pump, tubing, fittings...
THANK YOU FOR REPLY
yeah, i planing to buy water block's, pump fittings, tubing etc..
its not only about money, i am interested in this question, and i'd like to see in test, and after than upgrade it .
 
Nov 11, 2019
5
0
10
0
Heat capacity of water is 4184 Joules per degree C per kg.

[40 * 4184 * (80-25)]/600W = 15341s = 4.26 hours

That's a very rough estimate, as it assumes that no heat in the tank is being dissipated into the room. But the point is that a closed tank of water is not an efficient heat exchanger, meaning that eventually it's going to heat up at which point it won't be able to cool your system anymore.
TJ Hooker i think you are right ! maybe its a bad idea. but what if to use a Central heating radiator instead of big water tank, I seen how Linus did it once
 

Eximo

Titan
Herald
It might be fun for a bit, or you can dump ice in it for some fun overclocking runs. But radiators will work better long term. (Even then I usually see people put whole radiators in an ice bath rather than exposing the loop directly to the cooled water)

That or find yourself a swimming pool to use.

I've always wanted to try out a phase change cooler, but I saw der8auer has made one with automated overclocking on top of that.

My new silly idea is to see about a super critical fluid. Still be phase change, but also require silly levels of equipment for high pressure tolerance. Also it would probably boil away too quickly from a CPU or GPU to work properly.
 
Reactions: Zsoltluu

rubix_1011

Contributing Writer
Moderator
The large 'reservoir' would need to be designed something like the old Zalman Reserator (complete garbage cooler, and most failed).



Basically, it is a large cylinder that is also a finned heatsink for dissipation. Zalman thought it would be good to make these out of aluminum, but didn't stop to think that the other components might contain brass or copper, which led to large-scale galvanic corrosion.

Other solutions would be to purchase the same watercooling components with a radiator (such as a 360), remove the fans and place the radiator into an ice chest or cooler submerged in ice water. Many overclockers have done this, but ice needs to be constantly maintained, otherwise, once the water warms, it simply continues to get warmer and warmer, negating the benefit and actually suffering from the insulation properties of the beverage cooler.
 

USAFRet

Titan
Moderator
Mar 16, 2013
115,450
2,271
145,090
18,762
TJ Hooker i think you are right ! maybe its a bad idea. but what if to use a Central heating radiator instead of big water tank, I seen how Linus did it once
Yes, Linus may have done it.
As a demo for what is basically a TV show. With his large staff and free hardware.

A house size radiator, car size rad, or motorcycle size rad...gives you zero performance benefit over a good AIO or custom PC sized loop.
A good air cooler uses the thermal mass of the whole house air volume. Not a little fish tank size thing.

And the other major restriction is how fast you can remove the heat from a 25mmx25mm contact patch on the CPU.
Move all the liquid you want, but that contact patch isn't getting any bigger.
 
Reactions: Zsoltluu
Nov 11, 2019
5
0
10
0
Doubtful. Here's why.

What you are suggesting is a large watercooling reservoir. That's all. The tank would need to be metal or a substance that has the ability to dissipate thermal energy. Ambient room temperature will always need to be equal to or lower than the water inside the tank. It will also need to be protected from growth using some form of inhibitor or biocide. This simply cannot be 'just water'. It won't take long for the warmth of the cooling loop to provide a place for microbials to bloom. The tank also cannot be made of anything that is going to contribute to the introduction of galvanic corrosion.

Second, you need a pump and water blocks to be effective. A good CPU block is between $50-$80 (US dollars) while a good GPU block (full cover) is $100-$140 US dollars. A good pump like a DDC or D5 is going to run you another $65-$85 US dollars. You will need fittings, and even the most basic barb fittings are about $2.50 a piece...you will need at least 2 for every component. You will also need tubing - some of the cheapest is around $1.50 per linear foot.

This doesn't even touch the 'large reservoir' tank you are describing. Where will you source this? What size/capacity? This is very important, and here's why: it MUST be able to dissipate more thermal load in watts than the cooling loop components (CPU + GPU) are producing each second, otherwise, the temperature of the water will continue to rise higher and higher unless load is diminished.

I myself have a 9700k and an RTX 2080 and a fully watercooled loop for them. I can tell you that under load, those will put out some watts.

So, I would say that this idea is not very feasible nor is it done at a low cost, unless you can get much of the above for very, very, very low cost or free.

Do not steal.
THANK YOU FOR REPLY
it was a theory I was thinking about and I am aware of the costs and with necessarily parts .
 

ASK THE COMMUNITY

TRENDING THREADS