Jump to content

Passively cooled fish tank computer

Just looked up a similar looking container, dimension are about ~ 28 " L x 21" W x 18 " H

 

Idle temps are around 25-30C.

Container is sealed with a snapping top lid.

System on-off every day used throughout the whole day.

 

The 80-90C is hit with prime95 and with old clock speeds and most importantly, an extremely subpar ebay block (15$ ebay, fin thickness 1mm, no jet plate).

 

Also, i have used a smaller container before to test the feasibility of this build. about the same size as the one im using now but half the height.

 

I thought going to a larger size with double the water capacity and more surface area would help. But temps are identical.

 

So you hit some point where the temps equalize and heat given off from the surface of the tank will = heat being added to the system.

 

Edit: forgot to add, coolant is tap water with detergent in it. Been running it since the start of the build. Seems good with no bacterial growth. Only problem is, the EK block has very thin fins and dust gets into the container whenever i open the lid etc. and the block needs to be cleaned out like once a year maybe....(havent cleaned it yet flow hasn't decreased)

Link to comment
Share on other sites

Link to post
Share on other sites

Just looked up a similar looking container, dimension are about ~ 28 " L x 21" W x 18 " H

 

Idle temps are around 25-30C.

Container is sealed with a snapping top lid.

System on-off every day used throughout the whole day.

 

Also, i have used a smaller container before to test the feasibility of this build. about the same size as the one im using now but half the height.

 

I thought going to a larger size with double the water capacity and more surface area would help. But temps are identical.

 

It makes sense that the temperatures would be the same. With the lid on the water isn't able to evaporate so the only way the water can cool is by radiating heat from the sides of the bin. Do you know how long it takes for the water to go from room temperature to 30?

 

I don't know what your TDP is since you're underclocking/volting your chip but if your CPU and GPU are 250W total. The box is 28"x21"x18" = 45.8 gal. I doubt it's filled to the very top so I'll say 42 gal. 30C is 86F, which is warm but I was expecting the water to get up to 100F (~38C). If it's only getting to 30C, with a starting temperature of 20C, that's 9.4 hours. I guess the big unknown is the ebay CPU waterblock. 70C makes me uncomfortable, so if your chip is hitting 90C... :\

 

With the lid off (which sounds like a big no-no) that tub is ~4 square feet which radiates 155.6W which results in 95W going into the water. At that rate, to get from 20C up to 30C it should take 24.8 hours.

 

Are there any radiators or anything like that on the system? Exposed copper pipes?

Link to comment
Share on other sites

Link to post
Share on other sites

no exposed copper pipes but quite a long section of flexible plastic tubing.

The box lid isn't 100% sealed though and there's is evaporation still occurring. The lid of the box is clipped on but some air (albeit very little) can still go in and out. can't really give you time from 20 C to 30 C as the box is still warm to the touch the next morning.

I'm giving rough numbers here but I've had to refill the bin after about 2 months use. About 2-4 cm water height loss.

Also I still cannot stability test my cpu with prime 95 even though it is undervolted. It's just everything I do with it daily doesn't push it all that hard.

Link to comment
Share on other sites

Link to post
Share on other sites

One thing I have to say, you might get temperature stratification in the water tank if it is not churning.  This simply means hot water will stay on top and cold water will stay on bottom.  This can actually be beneficial in this case, as long as you are pulling the cold water from the bottom of the tank and depositing hot water on the top.  You can just extend the cold water pipe down from inside the tank to get this effect.  It might even be more advantageous to get a slower pump to aid in the stratification of the water.  (higher flow rates = more likely to 'churn' the water and equalize the temperatures from top to bottom).  Also, taller tanks tend to stratify better.  If you do end up putting a radiator on the tank, just do the opposite: pull hot water from the top and deposit cold water on the bottom.

 

For a macro-sized example of this, check out the image http://facilities.ucmerced.edu/central-plant.  That huge cylinder at the right of the picture that is dwarfing the 3 story building is filled with two million of gallons of water.  They cool down the water at night and use the cold water as an AC system during the day for 4 large buildings (not pictured).  They use the thermal stratification to keep the warm water from mixing with the cold water, meaning they don't need 2 separate tanks.

Link to comment
Share on other sites

Link to post
Share on other sites

Hey there. Welcome to the forums! Love the intricate planning. A pleasure to see. Hope it goes well for you.

Bleigh!  Ever hear of AC series? 

Link to comment
Share on other sites

Link to post
Share on other sites

Without getting into detail, people have tried this and it doesn't work. It's a stupid idea.

Intel Inside. Overweight guy in his 30's outside.

Link to comment
Share on other sites

Link to post
Share on other sites

Without getting into detail, people have tried this and it doesn't work. It's a stupid idea.

Evidence?

Everything you need to know about AMD cpus in one simple post.  Christian Member 

Wii u, ps3(2 usb fat),ps4

Iphone 6 64gb and surface RT

Hp DL380 G5 with one E5345 and bunch of hot swappable hdds in raid 5 from when i got it. intend to run xen server on it

Apple Power Macintosh G5 2.0 DP (PCI-X) with notebook hdd i had lying around 4GB of ram

TOSHIBA Satellite P850 with Core i7-3610QM,8gb of ram,default 750hdd has dual screens via a external display as main and laptop display as second running windows 10

MacBookPro11,3:I7-4870HQ, 512gb ssd,16gb of memory

Link to comment
Share on other sites

Link to post
Share on other sites

http://imgur.com/a/B1uP7

 

It works but its not ideal.

 

And I've also thought of the water circulation problem.

The bin has two holes drilled in it with fittings at the top and bottom left.

This allows for cross flow and you have full circulation.

Link to comment
Share on other sites

Link to post
Share on other sites

Warm water is going to be a cespool for algae and fungus.

 

You're not going to be able to support any hardware that's worth watercooling in the first place (a single 960 or something isn't worth the hassle of doing this)

 

You're better off running the lines to an array of radiators and calling it a day.  Similar to Whole Room's roof top array.

Workstation:  13700k @ 5.5Ghz || Gigabyte Z790 Ultra || MSI Gaming Trio 4090 Shunt || TeamGroup DDR5-7800 @ 7000 || Corsair AX1500i@240V || whole-house loop.

LANRig/GuestGamingBox: 9900nonK || Gigabyte Z390 Master || ASUS TUF 3090 650W shunt || Corsair SF600 || CPU+GPU watercooled 280 rad pull only || whole-house loop.

Server Router (Untangle): 13600k @ Stock || ASRock Z690 ITX || All 10Gbe || 2x8GB 3200 || PicoPSU 150W 24pin + AX1200i on CPU|| whole-house loop

Server Compute/Storage: 10850K @ 5.1Ghz || Gigabyte Z490 Ultra || EVGA FTW3 3090 1000W || LSI 9280i-24 port || 4TB Samsung 860 Evo, 5x10TB Seagate Enterprise Raid 6, 4x8TB Seagate Archive Backup ||  whole-house loop.

Laptop: HP Elitebook 840 G8 (Intel 1185G7) + 3080Ti Thunderbolt Dock, Razer Blade Stealth 13" 2017 (Intel 8550U)

Link to comment
Share on other sites

Link to post
Share on other sites

Warm water is going to be a cespool for algae and fungus.

 

You're not going to be able to support any hardware that's worth watercooling in the first place (a single 960 or something isn't worth the hassle of doing this)

 

You're better off running the lines to an array of radiators and calling it a day.  Similar to Whole Room's roof top array.

 

As previously discussed, a biocide is going to be used as per standard water cooling loops. The system is rarely used for more than a few hours at a time and I've already calculated the absolute worst-case scenario time to heat water based on the TDP of the components in my system assuming that the water is being stored in a perfectly insulated container. The goal with this build is to have a 100% passively cooled system where the whole-room system is supported by a massive array of fan-cooled radiators. If you can back up your claims that this won't work with examples of failures and/or mathematical formulas to prove this I'm all ears.

Link to comment
Share on other sites

Link to post
Share on other sites

Not sure why you're so upset that the rig won't be portable. It's left the house once in the 6 years since I built it. As far as how it looks, I've got that covered.

 

NRdqDAz.jpg

Link to comment
Share on other sites

Link to post
Share on other sites

So I think there's some misunderstanding with what my goal is. First, no oil. No oil anywhere. This is 100% water. Second, I'm not submerging my computer in water. Please see this extremely technical diagram I made to illustrate the idea.

 

YM2GByA.jpg

 

to avoid contamination you could submerse a 360mm rad in the water. have a water-water cooling system like a liquid cooled boat engine.

Sim Rig:  Valve Index - Acer XV273KP - 5950x - GTX 2080ti - B550 Master - 32 GB ddr4 @ 3800c14 - DG-85 - HX1200 - 360mm AIO

Quote

Long Live VR. Pancake gaming is dead.

Link to comment
Share on other sites

Link to post
Share on other sites

@FakeGamerGuy

 

Why don't you use a passive radiator to get rid of 100 - 200 watts, so you can extend the usage time with the same tank? It may also cool down the water when the PC is in idle.

Mineral oil and 40 kg aluminium heat sinks are a perfect combination: 73 cores and a Titan X, Twenty Thousand Leagues Under the Oil

Link to comment
Share on other sites

Link to post
Share on other sites

to avoid contamination you could submerse a 360mm rad in the water. have a water-water cooling system like a liquid cooled boat engine.

 

That's actually a really interesting idea. There would be some thermal efficiency lost and water around the radiator might come hotter than the rest of the tank resulting in further efficiency loss, but seems doable.

 

@FakeGamerGuy

 

Why don't you use a passive radiator to get rid of 100 - 200 watts, so you can extend the usage time with the same tank? It may also cool down the water when the PC is in idle.

 

That was discussed earlier. I want to build the system as it's described in the first post and monitor how effective it is first and add radiators if necessary to keep cost, complexity, and the number of failure points as low as possible. When I add my 2 GPUs to the loop it might become necessary.

 

The current idea is to make my own tank using acrylic, acrylic epoxy, and silicone caulk into a relatively narrow L-shape that runs around the back and side of my desk. Both edges are next to walls so I don't have to worry about it falling off. The advantage is similar or greater volume of water with a huge increase in surface area. In order to test the effects of making the tank taller or wider I created a little calculator that tells me:

  • Surface area
  • Volume
  • Time to heat the water to 100F
  • Watts radiated at the water surface

https://docs.google.com/spreadsheets/d/1fEDWcyzB3A_2rEVZqZw4AKSedAjhLA8o5X_MYqmIG3E/edit?usp=sharing

 

As mentioned earlier, none of this takes into account heat radiated from the sides of the tank, heat lost during warm-up, fluctuations in heat output of the devices, etc.

 

Does anyone know what the approximate correlation between water temperature and CPU temperature in a water-cooled system?

Link to comment
Share on other sites

Link to post
Share on other sites

I did some more research and, while I still haven't modeled the temperature increase over time including radiation loss, I have determined the heat radiated through the tank walls at specific temperatures. Using this page for the radiation formula and this page for the thermal conductivity of acrylic, I've determined that when the water is at 90F (32C) the sum of all radiation sources is ~302.9W and at 100F (~38C) the total radiation is 475.5W in a system with a peak TDP of 440W. If I'm only cooling my CPU, which is my short-term goal, equilibrium is achieved at 85F from wall radiation alone.  The spreadsheet in the previous post has been updated to reflect the new data. Estimating surface radiation and evaporation equilibrium should be hit at about 80F.

 

Here's the comparisons between thickness of the tank acrylic, water temperature, and total thermal radiation from all sources.

 

Wall thickness @ 90F/32C
1cm: 255 W

.75 cm: 303 W

.5 cm: 399 W

 

Wall thickness @ 100F/38C
1cm: 406 W

.75 cm: 475.5 W

.5 cm: 615 W

 

TL;DR Math says that this a totally functional system.

Link to comment
Share on other sites

Link to post
Share on other sites

I did some more research and, while I still haven't modeled the temperature increase over time including radiation loss, I have determined the heat radiated through the tank walls at specific temperatures. Using this page for the radiation formula and this page for the thermal conductivity of acrylic, I've determined that when the water is at 90F (32C) the sum of all radiation sources is ~302.9W and at 100F (~38C) the total radiation is 475.5W in a system with a peak TDP of 440W. If I'm only cooling my CPU, which is my short-term goal, equilibrium is achieved at 85F from wall radiation alone.  The spreadsheet in the previous post has been updated to reflect the new data. Estimating surface radiation and evaporation equilibrium should be hit at about 80F.

 

Here's the comparisons between thickness of the tank acrylic, water temperature, and total thermal radiation from all sources.

 

Wall thickness @ 90F/32C

1cm: 255 W

.75 cm: 303 W

.5 cm: 399 W

 

Wall thickness @ 100F/38C

1cm: 406 W

.75 cm: 475.5 W

.5 cm: 615 W

 

TL;DR Math says that this a totally functional system.

 

did you consider the temperature of the exterior of the tank? your math would suggest a 0c degree room.

Sim Rig:  Valve Index - Acer XV273KP - 5950x - GTX 2080ti - B550 Master - 32 GB ddr4 @ 3800c14 - DG-85 - HX1200 - 360mm AIO

Quote

Long Live VR. Pancake gaming is dead.

Link to comment
Share on other sites

Link to post
Share on other sites

did you consider the temperature of the exterior of the tank? your math would suggest a 0c degree room.

 

Yes. Assuming the water is 100F (37.8C) and a 0.75cm thick acrylic wall, we get the following scenarios:

Ambient 32F/0C: 0.2 * 0.588 m^2 * (37.8 - 0) / 0.0075 = 592.704 watts

Ambient 68F/20C: 0.2 * 0.588 m^2 * (37.8 - 20) / 0.0075 = 279.104 watts

 

Both of these figures are from radiation through the walls alone. At 100F there is another 196.6 watts of radiation from evaporation and surface radiation. Pardon the 0.2 watt difference between the above calculations and the spreadsheet linked earlier; I rounded some figures and it didn't.

Link to comment
Share on other sites

Link to post
Share on other sites

Yes. Assuming the water is 100F (37.8C) and a 0.75cm thick acrylic wall, we get the following scenarios:

Ambient 32F/0C: 0.2 * 0.588 m^2 * (37.8 - 0) / 0.0075 = 592.704 watts

Ambient 68F/20C: 0.2 * 0.588 m^2 * (37.8 - 20) / 0.0075 = 279.104 watts

 

Both of these figures are from radiation through the walls alone. At 100F there is another 196.6 watts of radiation from evaporation and surface radiation. Pardon the 0.2 watt difference between the above calculations and the spreadsheet linked earlier; I rounded some figures and it didn't.

 

Be careful! You only callculated the the thermal resistance of the acrylic itself. But you also have an transfer resistance form the water to the acrylic (smal, no problem) and from the acrylic to the air (big). You will have to add an other resistance of ~0.17 m^2*K/W. Your dissipiation is to high, because you missed one part of the formula.

 

I can back it up with a personal experience:

- Heat sink with 1.498m^2 surface area made of black annodized aluminium

- power dissipated: 180 watts

- temperature rise over ambient: 25°C

 

-> thermal resistance = 0.139 K/W => 0.208 m^2*K/W

 

The value of 0.17 I have from the internet and I calculated 0.2 by mayself. You see the "real" value is something along those lines. Add this to you calculation and redo the math again.

Mineral oil and 40 kg aluminium heat sinks are a perfect combination: 73 cores and a Titan X, Twenty Thousand Leagues Under the Oil

Link to comment
Share on other sites

Link to post
Share on other sites

to avoid contamination you could submerse a 360mm rad in the water. have a water-water cooling system like a liquid cooled boat engine.

 

Been there done that.

 

The conclusion is to use a water-water heat exchanger instead. 

 

They don't tell you that "all copper and brass" doesn't mean the case isn't made from steel:

 

20150101_232551.jpg

Workstation:  13700k @ 5.5Ghz || Gigabyte Z790 Ultra || MSI Gaming Trio 4090 Shunt || TeamGroup DDR5-7800 @ 7000 || Corsair AX1500i@240V || whole-house loop.

LANRig/GuestGamingBox: 9900nonK || Gigabyte Z390 Master || ASUS TUF 3090 650W shunt || Corsair SF600 || CPU+GPU watercooled 280 rad pull only || whole-house loop.

Server Router (Untangle): 13600k @ Stock || ASRock Z690 ITX || All 10Gbe || 2x8GB 3200 || PicoPSU 150W 24pin + AX1200i on CPU|| whole-house loop

Server Compute/Storage: 10850K @ 5.1Ghz || Gigabyte Z490 Ultra || EVGA FTW3 3090 1000W || LSI 9280i-24 port || 4TB Samsung 860 Evo, 5x10TB Seagate Enterprise Raid 6, 4x8TB Seagate Archive Backup ||  whole-house loop.

Laptop: HP Elitebook 840 G8 (Intel 1185G7) + 3080Ti Thunderbolt Dock, Razer Blade Stealth 13" 2017 (Intel 8550U)

Link to comment
Share on other sites

Link to post
Share on other sites

The current idea is to make my own tank using acrylic, acrylic epoxy, and silicone caulk into a relatively narrow L-shape that runs around the back and side of my desk. Both edges are next to walls so I don't have to worry about it falling off. The advantage is similar or greater volume of water with a huge increase in surface area. In order to test the effects of making the tank taller or wider I created a little calculator that tells me:

 

If you want to avoid a gigantic point for failure, don't try to make your own tank.  Premade aquarium, sterrilite container, garbage can are the best ways to hold water.  Especially for this much water that will destroy flooring and drywall if it fails.

 

Also, silicone sealant can't be used below the waterline.

Workstation:  13700k @ 5.5Ghz || Gigabyte Z790 Ultra || MSI Gaming Trio 4090 Shunt || TeamGroup DDR5-7800 @ 7000 || Corsair AX1500i@240V || whole-house loop.

LANRig/GuestGamingBox: 9900nonK || Gigabyte Z390 Master || ASUS TUF 3090 650W shunt || Corsair SF600 || CPU+GPU watercooled 280 rad pull only || whole-house loop.

Server Router (Untangle): 13600k @ Stock || ASRock Z690 ITX || All 10Gbe || 2x8GB 3200 || PicoPSU 150W 24pin + AX1200i on CPU|| whole-house loop

Server Compute/Storage: 10850K @ 5.1Ghz || Gigabyte Z490 Ultra || EVGA FTW3 3090 1000W || LSI 9280i-24 port || 4TB Samsung 860 Evo, 5x10TB Seagate Enterprise Raid 6, 4x8TB Seagate Archive Backup ||  whole-house loop.

Laptop: HP Elitebook 840 G8 (Intel 1185G7) + 3080Ti Thunderbolt Dock, Razer Blade Stealth 13" 2017 (Intel 8550U)

Link to comment
Share on other sites

Link to post
Share on other sites

 

Been there done that.

 

The conclusion is to use a water-water heat exchanger instead. 

 

They don't tell you that "all copper and brass" doesn't mean the case isn't made from steel:

 

You could have added a block of zinc into the water. I will act as an victim electrode (not sure this is the rigth translation) and get's eroded over time. As long as you renew it often enougth your steel doesn't rust.

If your car has rims made out of steel you will (should) finde a smal bock of zinc attached somewhere.

 

post-216771-0-49435400-1434541183_thumb.

Mineral oil and 40 kg aluminium heat sinks are a perfect combination: 73 cores and a Titan X, Twenty Thousand Leagues Under the Oil

Link to comment
Share on other sites

Link to post
Share on other sites

You could have added a block of zinc into the water. I will act as an victim electrode (not sure this is the rigth translation) and get's eroded over time. As long as you renew it often enougth your steel doesn't rust.

If your car has rims made out of steel you will (should) finde a smal bock of zinc attached somewhere.

 

attachicon.gifauswuchtgewichte.jpg

 

LOL I don't suggest ruining the balance of your wheels to stop rust from a hacked-together water cooler.

Link to comment
Share on other sites

Link to post
Share on other sites

I didn't really care about the case rusting (looked kinda mad max style and was a bit of a "how far will it go?" experiment), I was worried about the lead solder they probably use to put the fins together failing and then having a huge contamination problem with my desktop loop.  So I switched to stacked plate heat exchangers that are stainless and copper brazed with 4000W heat capacity at 1C:  tim-allen-grunt-o.gif HOHOHOHOHO

Workstation:  13700k @ 5.5Ghz || Gigabyte Z790 Ultra || MSI Gaming Trio 4090 Shunt || TeamGroup DDR5-7800 @ 7000 || Corsair AX1500i@240V || whole-house loop.

LANRig/GuestGamingBox: 9900nonK || Gigabyte Z390 Master || ASUS TUF 3090 650W shunt || Corsair SF600 || CPU+GPU watercooled 280 rad pull only || whole-house loop.

Server Router (Untangle): 13600k @ Stock || ASRock Z690 ITX || All 10Gbe || 2x8GB 3200 || PicoPSU 150W 24pin + AX1200i on CPU|| whole-house loop

Server Compute/Storage: 10850K @ 5.1Ghz || Gigabyte Z490 Ultra || EVGA FTW3 3090 1000W || LSI 9280i-24 port || 4TB Samsung 860 Evo, 5x10TB Seagate Enterprise Raid 6, 4x8TB Seagate Archive Backup ||  whole-house loop.

Laptop: HP Elitebook 840 G8 (Intel 1185G7) + 3080Ti Thunderbolt Dock, Razer Blade Stealth 13" 2017 (Intel 8550U)

Link to comment
Share on other sites

Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now


×