three stage cooling oil and water chilled peltier Hobby experiment only

[monoceros21]

so first i am a mechanic not a pc builder and this is not a pc we a discussing this is a cooling system. this is a hobby/experiment out of my curiosity because I've noticed something in almost all mineral oil pc threads. no one seems to understand that oil is best cooled with water( and to the one who said water is used to cool cars not oil you are wrong water cools the oil) through liquid to liquid heat exchange, and due to the fact that oil can cover more surface area than a water block via heatsink the thermal conductivity factor can be made up for.

that aside the origin of this experiment is i do want a mineral oil pc because of the cool and wtf factor, but why not make it work well to.

the idea now. every mineral oil pc i have seen uses air to liquid heat exchange to keep the oil maintained but because of oils ability to retain the heat it does absorb this is not effective. so my idea (which credit to 4ryan6 to posting the peltier build that is stage 3 of this) is as follows.

first oil cools the pc cycled through a liquid to liquid heat exchanger, second chilled water through the exchanger as well in a cross flow configuration. this means that the coldest water flows in where the coldest oil flows out and the same for the hot sides. third chill the water with peltiers. fyi im not talking about water really more like non conductive coolant with a real low freezing point. a forth stage could be considered to be keeping the hot side of the peltiers as cool as possible since longer runtimes would be better for the cold water and relay thermostats are already planed in this build

the hope is that if the pc i put in this build were to raise the oil to say 50c with out the chilled water. the chilled water could bring it down to in a dream world 25 in reality maybe 30 to 35 c as the maintained temp.

please if your responses are why would you do this. your and idiot or any other trolling keep it to your self i am well aware of the draw backs namely the fact that the plans i already have have me consuming 1 kilo watt an hour when in full operation.

what i am looking for in posting this is constructive criticisms, potential idea on how you would do it and maybe that silent person who has already done this and how it worked and what they would do different. i do have a full build planned out but first im curious what ideas others have and will give more details if interested.
UPDATE: BUILT AND RUNNING.
but sadly with a leak in the bottom of the heat exchanger. it didnt have an impact on cpu temps ove water cooling i have a 6900k overclocked to 4.4Ghz and thats as far as i can get it. stress testing puts it at 64C. gpu temps however it did awsome i have yet to get the gpus over 44C. 2 asus 1080ti strix over clocked to 1757 boost 3.0 to 2063 memory at 11120. and every thing else on the board has seen 15 to 20 degree drops. a samsung 960 pro 2Tb went from 50c to 29.
8 avexir green teslas 2800 c15 are oced to 3000 and went from between 54 and 57 to 31 and 36. all this while the oil never gets over 29C its self.
now if i had enough cooling overhead on the cold water i was going to remove the air coolers and go to water blocks for the cpu and gpu. how ever the cold water get heated to between 18 and 24C. before i put any load it was sitting at 1 to 4C based on ambient. and non of this does any thing to prevent a massive heat output in to the room as i went with the equivalent of 17 140mm radiators to keep the hot water within 30 degrees of 0 the hot water goes from 34 when the thermostat shuts the sytem of to 29 the cycles the peltiers back on. with that amount of 34 degree surface area the system has a huge thermal and physical foot print. the case was built out of aluminum ex and sheet metal and is 800mm by 800mm by 600mm. ill post picture when i can but back to waterblocks. the oil enter thru a port i drilled in the bottom of the plexi tank and exits 3/4 the way up just behind the cpu sucking the hot oil away from the cpu backing area for the mobo. what i may do in the end is change from air coolers transfering heat to the oil is i might just use the oil as an insulator to prevent condensation from forming on the cold water lines by removing the heat exchanger alltogether and use the bottom port to enter the tank under the oil thus keeping air off the lines. i would have loved to do this and keep the heat exchanger but its not working well enough.... it might if i switch ti a 14.4 powersupply since im using 26 amp peltiers at a 20 amps 12vdc off of a ax1200i, but i like being able to monitor the draw off the power supply the way i am.
as for the leak its very slow and manageable since i made a removable drip pan at the bottom of the case and designed the case floor with a bend to flow all liquids to one spot.
so any suggestions on which way i should go first up the power supply or go to mineral oil insulating the cold lines? the tank is holing up very well and has its corners caged by the aluminum ex and the holes have bits power bulkhead pass thrus with double packings to keep it from leaking under the preasure of the oil so im quite confident the will hold up but if cold water is put thru those ports it might shrink the packings and then leak? or even then the cold water might still get heated before it hits the cpu since the oil will then be even hotter. what do you all think?

 

[rubix_1011]

In a car, oil is the lubricant and while it does absorb heat, the cooling journals are meant to absorb heat from the cast components in which the coolant runs through. Otherwise, you'd have a heat exchanger with oil/water being maintained rather than water running throughout the block and head(s) and back to the radiator. Essentially, your block and engine heads are just big water blocks like you'd find in a CPU cooling loop. Oil is essentially your thermal paste.

I think what you're saying here is that the ability for oil to successfully transfer heat from friction points (connecting rods, crank etc) to the heads and block to then be transferred to the water journals to be cooled, which yes, that has to be the case because you cannot have coolant in direct contact with components requiring lubrication under high stresses. We all know what happens to coolant when it mixes with oil in an engine - and it isn't good. However, oil isn't cooling your engine - if you add a quart of oil over the fill line, your engine doesn't run cooler. If your engine is a quart low, it could run hotter, but only because of the additional friction and lack of cooling time the lubricant has, even if there is an oil cooler. Likewise, an engine full of dirty oil will run hotter than an engine of clean oil due to differences of friction and fluid density. If oil was doing that much cooling, your radiator would be an oil/liquid+ liquid/air exchanger rather than liquid/air exchanger with your coolant.

In your scenario, is this a mineral oil submersion or are you planning to pump the mineral oil through a cooling loop and then exchange that with a liquid cooling loop?

Also consider that the specific heat of water is greater than that of light oil: http://www.engineeringtoolbox.com/specific-heat-fluids-d_151.html

 

[monoceros21]

But isnt it. Water cools the block. The block cools the oil or a seperate oil cooler to cool it even more and the hottest parts of the engine the oil lubs and takes the heat away. The reason you engine gets hotter when the oils low is your recirculating the oil faster than it can maintain stability. If oil doesn't stay cool it will break down. Its all heat transfer. Moving parts send it to the oil. The oil to the block the block to the water the water to the radiator. The radiator to the air.
Anyway oil submersion is my plan. My reasoning behind it is just looks cool but i want it to work as good as possible. And to do that i believe a liquide to liquide heat exchange with chilled coolant on the other side instead of an air to liquid radiator. For the cpu and the fact that water does have better thermal values, about 5 times better for conductivity , i beleive a all copper heat sink about 5 times the size of a cpu water block and a pump jetting out the cold oil near the cpu fan would compensate for the differences in thermal properties.
Now im an aircraft mechanic and could post some links about oil concerning aircraft and cooling but im on my phone right now and that would just be a pain so later i will. And cars are different than planes. But oil is a secondary coolant in high altitude planes where ever fan or ram air cannot reach, but it is also a fuel warmer which is how the oil is cooled on these planes.

 

[rubix_1011]

I would imagine that the reason oil is used to cool plane hardware is due to the freezing points (or potential thereof) of water based coolants where oil would not encounter freezing at high altitudes and very low temperatures. While this would work, it works because it has to and the potential for failing is far too great to use water that could freeze and cause a fatal failure.

Automobile engine blocks and heads all have water and oil journals. If you get a cracked head due to overheating due to low coolant level, and then you add coolant, you end up with water in your oil. I've seen this happen first hand on engines I've pulled heads and replaced. Likewise, if you have a warped head, you'll end up with oil/water mix as well due to the oil and water journals basically inter-mixing.

Not all cars have oil coolers, most do not actually. Most cars with automatic transmissions have transmission coolers that are trans oil to air via radiator that is mounted either in front or behind of the coolant radiator.

I understand what you are saying about oil being between the heat producing components and the water cooled section - this is the same thing that happens in a normal watercooling loop - water does not directly contact the electrical components producing heat. Heat spreaders and thermal paste act as an interface between them. Oil in an engine is primarily used for lubrication, not temperature control. As heat in the engine is produced, it radiates outward toward the water journals since these are colder than the load temperature the engine is creating. You are correct in that engine oil does assist in helping the engine dissipate thermal load (like thermal paste) but it isn't really doing the heavy lifting of the cooling system which is handled by the water pump, coolant and radiator.

 

[monoceros21]

I would imagine that the reason oil is used to cool plane hardware is due to the freezing points (or potential thereof) of water based coolants where oil would not encounter freezing at high altitudes and very low temperatures. While this would work, it works because it has to and the potential for failing is far too great to use water that could freeze and cause a fatal failure.

Automobile engine blocks and heads all have water and oil journals. If you get a cracked head due to overheating due to low coolant level, and then you add coolant, you end up with water in your oil. I've seen this happen first hand on engines I've pulled heads and replaced. Likewise, if you have a warped head, you'll end up with oil/water mix as well due to the oil and water journals basically inter-mixing.

Not all cars have oil coolers, most do not actually. Most cars with automatic transmissions have transmission coolers that are trans oil to air via radiator that is mounted either in front or behind of the coolant radiator.

I understand what you are saying about oil being between the heat producing components and the water cooled section - this is the same thing that happens in a normal watercooling loop - water does not directly contact the electrical components producing heat. Heat spreaders and thermal paste act as an interface between them. Oil in an engine is primarily used for lubrication, not temperature control. As heat in the engine is produced, it radiates outward toward the water journals since these are colder than the load temperature the engine is creating. You are correct in that engine oil does assist in helping the engine dissipate thermal load (like thermal paste) but it isn't really doing the heavy lifting of the cooling system which is handled by the water pump, coolant and radiator.

exactly. but how it cools by moving the heat like that is what i'm trying to imitate. oil is oil. coolant would be the fuel which becomes cold because the fuel is in the wing with the cold air that comes with high altitude passing above and below. the coolant if 4ryan6's peltier posts are correct can get to well below 0 which with the oil being warmed by the pc flowing cross flow to this should take the oil heat pretty effectivly just like the fuel in a plane or the journals in a car. oil covering a larger surface area of the pc than water ever could bring the heat to a preferably copper L to L heat exchanger (still soucing a reasonable one) that has the coolant chilled by a peltier running through it should make the system as efficent as water cooling and i get my pc in an aquarium to freak people out lol.
there is however one major concern. you are right journals crack and so do liquid to liquid heat exchangers. while i would love to take that cold exit oil from the exchanger and pump it right out on to the cpu and heatsink heat pipes what if it leaks. and since there is no such thing among the pc coolants as a non conductive coolant that is stable at the temps im thinking id need as far as i can tell. i could do a double wall exchanger, but one they are extremly expensive and two can't find in copper yet. the surface area range i've been looking at is 1.5 m to 2 meters basicly 18 in by 3 in 40 to 60 plate. and double walls in this range are over 1000 us. any ideas on that?

 

[Outlander_04]

There is no liquid to liquid cooling of the oil in a car .

The water cools the metal , some of the heat transfers from oil to the metal and is then removed via the water to the metal in the radiator to the air passing through it .

Just use oil > metal > air
the water stage is nothing more than a delay in finally getting rid iof the heat from the hot spots in the pc

If yo

 

[monoceros21]

There is no liquid to liquid cooling of the oil in a car .

The water cools the metal , some of the heat transfers from oil to the metal and is then removed via the water to the metal in the radiator to the air passing through it .

Just use oil > metal > air
the water stage is nothing more than a delay in finally getting rid iof the heat from the hot spots in the pc

If yo

but air is not as efficient as water. and the point is to try and get the same efficiency as water cooling in an aquarium. to use air would only make it as efficient as air since air has less desirable thermal properties. air in the end would be involved since i am using peltiers but the acceptable hi temp for those to operate is 125c vs the cpu's 80c so i have more room for less efficiency on the hot side of the peltier. the way i expect to be able to transfer the heat is heat sink> oil>copper L toL heat exchange with chilled coolant. and then in extension because laws of thermal dynamics to cool the chilled coolant i generated heat cold side of the peltier>hot side> distilled water>copper radiator> air.

 

[GlomGubben]

I'm tinkering with almost the same idea, but I was thinking of only cooling the oil with peltier.

My thought was to buy a cpu cooler like lets say Noctua NH-D15 for the cpu, and then using multiple Peltiers to cool down one or two rads submerged in the oil. or just a powerful pump and pump the oil to the waterblocks beeing cooled by the peltiers and back.

Have you made any headway with your idea?

 

[monoceros21]

I'm tinkering with almost the same idea, but I was thinking of only cooling the oil with peltier.

My thought was to buy a cpu cooler like lets say Noctua NH-D15 for the cpu, and then using multiple Peltiers to cool down one or two rads submerged in the oil. or just a powerful pump and pump the oil to the waterblocks beeing cooled by the peltiers and back.

Have you made any headway with your idea?

I'm still gathering the parts. i found i had to set up three relay boxes for three temp controllers since like you i will be using multiple peltiers and the controllers cant relay the amps required for more than one peltier i found some automotive relay boxes on del city. i would advise you get a temp controller as well for safety. on for hot side of the peltier, one for oil you don't want it to get to cold other wise you would need an insane amount of head pressure to push the cold oil, and if you go the cold water thru rad route then on for that as well so you dont freeze your coolant. i've got most of the system parts i just need to build a frame and the oil tank is coming in tomorrow. im issues getting the controller i wanted though. i had ordered xh-w1403 but its been three months and 2 separate orders and no luck. way to go china. looks like i'll have to got with the bulkier mh1210
as for the rad in oil with cold water going thru it. that is essentially exactly the same idea, and might serve you better as the oil thru a water block might not flow in there long enough to get cold, and if you slow down the flow it wont flow fast enough to effect the rest of the oil in the tank. i didnt want to do a rad in the oil because if the rad leaks the coolant might get too close to the components and fry it. remember water displaces oil and most coolants are distilled water as a main ingredient. i should be testing in late September or October the amount of fitting i needed for this is insane though. i either needed 12 of each type of qd which with koolance was 600 us. or 30 compression 7 ball valves and 4 more quad manifolds for 300 us i think the choice was obvious there. if you go with less safety's than me you want need as much though.

 

[GlomGubben]

I'm still gathering the parts. i found i had to set up three relay boxes for three temp controllers since like you i will be using multiple peltiers and the controllers cant relay the amps required for more than one peltier i found some automotive relay boxes on del city. i would advise you get a temp controller as well for safety. on for hot side of the peltier, one for oil you don't want it to get to cold other wise you would need an insane amount of head pressure to push the cold oil, and if you go the cold water thru rad route then on for that as well so you dont freeze your coolant. i've got most of the system parts i just need to build a frame and the oil tank is coming in tomorrow. im issues getting the controller i wanted though. i had ordered xh-w1403 but its been three months and 2 separate orders and no luck. way to go china. looks like i'll have to got with the bulkier mh1210
as for the rad in oil with cold water going thru it. that is essentially exactly the same idea, and might serve you better as the oil thru a water block might not flow in there long enough to get cold, and if you slow down the flow it wont flow fast enough to effect the rest of the oil in the tank. i didnt want to do a rad in the oil because if the rad leaks the coolant might get too close to the components and fry it. remember water displaces oil and most coolants are distilled water as a main ingredient. i should be testing in late September or October the amount of fitting i needed for this is insane though. i either needed 12 of each type of qd which with koolance was 600 us. or 30 compression 7 ball valves and 4 more quad manifolds for 300 us i think the choice was obvious there. if you go with less safety's than me you want need as much though.

Good to hear your still going forward

Yeah I have been looking at temp controllers on eBay, but will buy that later after I've gotten the system up and running and i'm certain it will work ^^ just gonna buy some Pulse Width Modulators and control them manually first.

Will be monitoring the temps, but I want it cold I'm aiming for the oil to be at least below 10c.
but yeah I am a little concerned for the oil thickening when the temp drops, but I'm hoping the pump these guys used will be sufficient: http://
I just need to find some one that sells it, or a similar pump ^^

Will most probably avoid using water. It may become necessary for cooling the peltiers, but then i'm not gonna make it possible for a water leak to get inn the tank ^^

The water blocks I'm gonna use will be something like this :

Will have the water blocks just a couple of centimeters apart, and if that isn't enough i'll have the oil go through a radiator with some über silent fans on

I see from those prices that safety will not be my No.1 priority right away x)

But it seems that I've hit a bit of a snag, was going to use a empty fish tank but since my graphic card is 305mm long and the fishtank has an inside height of 295mm I might have to buy another

Or I could buy a new cooler for it which would shorten it by 10-30mm but i'm guessing that would cost more than a new fishtank or whatever ^^

I'll maybe make a new thread about the build and progress, will post a link if I do =)

 

[monoceros21]

I'm still gathering the parts. i found i had to set up three relay boxes for three temp controllers since like you i will be using multiple peltiers and the controllers cant relay the amps required for more than one peltier i found some automotive relay boxes on del city. i would advise you get a temp controller as well for safety. on for hot side of the peltier, one for oil you don't want it to get to cold other wise you would need an insane amount of head pressure to push the cold oil, and if you go the cold water thru rad route then on for that as well so you dont freeze your coolant. i've got most of the system parts i just need to build a frame and the oil tank is coming in tomorrow. im issues getting the controller i wanted though. i had ordered xh-w1403 but its been three months and 2 separate orders and no luck. way to go china. looks like i'll have to got with the bulkier mh1210
as for the rad in oil with cold water going thru it. that is essentially exactly the same idea, and might serve you better as the oil thru a water block might not flow in there long enough to get cold, and if you slow down the flow it wont flow fast enough to effect the rest of the oil in the tank. i didnt want to do a rad in the oil because if the rad leaks the coolant might get too close to the components and fry it. remember water displaces oil and most coolants are distilled water as a main ingredient. i should be testing in late September or October the amount of fitting i needed for this is insane though. i either needed 12 of each type of qd which with koolance was 600 us. or 30 compression 7 ball valves and 4 more quad manifolds for 300 us i think the choice was obvious there. if you go with less safety's than me you want need as much though.

Good to hear your still going forward

Yeah I have been looking at temp controllers on eBay, but will buy that later after I've gotten the system up and running and i'm certain it will work ^^ just gonna buy some Pulse Width Modulators and control them manually first.

Will be monitoring the temps, but I want it cold I'm aiming for the oil to be at least below 10c.
but yeah I am a little concerned for the oil thickening when the temp drops, but I'm hoping the pump these guys used will be sufficient: http://
I just need to find some one that sells it, or a similar pump ^^

Will most probably avoid using water. It may become necessary for cooling the peltiers, but then i'm not gonna make it possible for a water leak to get inn the tank ^^

The water blocks I'm gonna use will be something like this :

Will have the water blocks just a couple of centimeters apart, and if that isn't enough i'll have the oil go through a radiator with some über silent fans on

I see from those prices that safety will not be my No.1 priority right away x)

But it seems that I've hit a bit of a snag, was going to use a empty fish tank but since my graphic card is 305mm long and the fishtank has an inside height of 295mm I might have to buy another

Or I could buy a new cooler for it which would shorten it by 10-30mm but i'm guessing that would cost more than a new fishtank or whatever ^^

I'll maybe make a new thread about the build and progress, will post a link if I do =)

check out us.misumi-ec.com if you register you get 30% off for 30 days you can get alot from them including an acrylic case made to spec http://us.misumi-ec.com/vona2/mech/M1400000000/M1411000000/M1411040000/
also for water blocks check out custom thermal electric. better quality than the blue amazon ones.
my solution for the water is i will be drilling a hole in the acrylic and using a passthru to port the oil in to the bottom of the tank. this way if there is an internal leak in my heat exchanger the water will go straight to the bottom so i wont have to worry about using non conductive fluid. then to get that cold oil mixed in with the hot in the tank ill be positioning fans in the tank. the heat exchanger i got is way over kill. https://www.dudadiesel.com/choose_item.php?id=HX3660

 

[monoceros21]

UPDATE: BUILT AND RUNNING.
but sadly with a leak in the bottom of the heat exchanger. it didnt have an impact on cpu temps ove water cooling i have a 6900k overclocked to 4.4Ghz and thats as far as i can get it. stress testing puts it at 64C. gpu temps however it did awsome i have yet to get the gpus over 44C. 2 asus 1080ti strix over clocked to 1757 boost 3.0 to 2063 memory at 11120. and every thing else on the board has seen 15 to 20 degree drops. a samsung 960 pro 2Tb went from 50c to 29.
8 avexir green teslas 2800 c15 are oced to 3000 and went from between 54 and 57 to 31 and 36. all this while the oil never gets over 29C its self.
now if i had enough cooling overhead on the cold water i was going to remove the air coolers and go to water blocks for the cpu and gpu. how ever the cold water get heated to between 18 and 24C. before i put any load it was sitting at 1 to 4C based on ambient. and non of this does any thing to prevent a massive heat output in to the room as i went with the equivalent of 17 140mm radiators to keep the hot water within 30 degrees of 0 the hot water goes from 34 when the thermostat shuts the sytem of to 29 the cycles the peltiers back on. with that amount of 34 degree surface area the system has a huge thermal and physical foot print. the case was built out of aluminum ex and sheet metal and is 800mm by 800mm by 600mm. ill post picture when i can but back to waterblocks. the oil enter thru a port i drilled in the bottom of the plexi tank and exits 3/4 the way up just behind the cpu sucking the hot oil away from the cpu backing area for the mobo. what i may do in the end is change from air coolers transfering heat to the oil is i might just use the oil as an insulator to prevent condensation from forming on the cold water lines by removing the heat exchanger alltogether and use the bottom port to enter the tank under the oil thus keeping air off the lines. i would have loved to do this and keep the heat exchanger but its not working well enough.... it might if i switch ti a 14.4 powersupply since im using 26 amp peltiers at a 20 amps 12vdc off of a ax1200i, but i like being able to monitor the draw off the power supply the way i am.
as for the leak its very slow and manageable since i made a removable drip pan at the bottom of the case and designed the case floor with a bend to flow all liquids to one spot.
so any suggestions on which way i should go first up the power supply or go to mineral oil insulating the cold lines? the tank is holing up very well and has its corners caged by the aluminum ex and the holes have bits power bulkhead pass thrus with double packings to keep it from leaking under the preasure of the oil so im quite confident the will hold up but if cold water is put thru those ports it might shrink the packings and then leak? or even then the cold water might still get heated before it hits the cpu since the oil will then be even hotter. what do you all think?