Water-cooling a heat pipe?

[Spindel]

Ok just get this thing off the table first. 

I haven't build a PC since early 2000's, and when I dit that all I used was basic aluminium/copper block air coolers. I do enjoy watching channels like LTT etc and it gives me an ich to build a PC but it would be kind of a waste since I would never use it. 

 

So the following post is a thing that I was just thinking of, from a clear thermodynamical stand point.

 

Shouldn't there be a gain from water-cooling a heat pipe instead of using the standard blocks (with or without fins) used for water-cooling?

 

The base for this is that a heat pipe works internally by evaporation and should always be better at transporting away the heat than a slab of solid copper (or any metal), and to dissipate this heat using water against the heat pipe should be much more efficient than having air (with a fin array).

If you take in example a Noctua NH-U12 and strip away the fins and apply a waterflow over the hot side of the heat pipes you should in theory be able to achieve a much higher heat flow than both the standard NH-U12 set up and than a standard water block. 

 

This should be even better if you can flatten a side of the heat pipes and have them contact directly against the CPU/IHS to reduce the amount of thermal resistance between the heat pipe and CPU.

 

Has anyone tried this? (and I mean in a proper build, not just building it like the youtube channel Major Hardware, that just made a water enclosure around an air cooler).  

 

If this gave an idea to anyone to try and make a device like this please share benchmarks to how it performs. 

 

Maybe this is something LMG should try to make a proper build of this in their series of home made CPU coolers.

 

[adm0n]

There is a problem to your theory.

Heat transfer scales with surface area as well as the thermal conductivity.

 

Thermal conductivity between two touching metals is a lot higher than between air/water and a metal. So first distributing the heat to a larger surface and then dissipating it, is the way to go.

[The Flying Sloth]

I'm not going to lie, I've had the same thought about running water through the pipes on an air cooler but here's what you missed, as previously stated thermal transfer scales with surface area, there's a lot more surface area in a water block with a microfin array than there is in the smooth edges of a head pipe, not to mention the size tifference between a heat pipe and the thermal contact zone in a water block.

I still want to do it at some point for fun but I understand that it's not a practical solution.

[SupaKomputa]

 

[Enderman]

Ok so your logic is that watercooling a CPU heatsink with its heatpipes and fins would work better than air, this is true.

It would also completely defeat the watercooling advantages of being far more compact and looking better.

 

 

However, it would not perform better than watercooling for several reasons.

 

1) Yes heatpipes have extremely high thermal conductivity, however having CPU -> 3mm  of copper -> fins has a lower total conductivity than CPU -> 150mm of heatpipes -> fins

That's because heat transfer depends on distance, in W/mK m means meters, and 3mm of 300W/mK copper is less restrictive than 150mm of 3000W/mK heatpipes.

 

2) Liquid cooling uses the flow of water to move the heat away rather than conductivity through a metal.

While a heatpipe has a fixed distance and speed it can carry away the energy, water flow can just go farther and farther by increasing head pressure, and can go faster and faster by increasing flow rate.

So essentially, although a different form of transmission, water can move energy MUCH faster than a heatpipe can.

 

3) Heatpipes are still made of copper, so even having direct-contact on the CPU (which a lot of air coolers do if you look at the pics) the heat still has to go through copper and enter the heatpipe. Heatpipes need surface area to have the heat enter and leave them, and need a large enough temperature differential to evaporate and condense the water inside but also not too high or else the water will not condense.

 

4) Watercooling has the advantage of having microfins right at the CPU, only a few mms away. Since the CPU already has such a small area, the problem is not really the conductivity of the copper or heatpipes that are touching it, it is the temperature of the object that is touching it. By having the water cool the microfins right there, the only thermal resistance between the water and the CPU is the thin copper plate of the bottom of the waterblock. Much lower than long heatpipes and then a stack of aluminum fins. Heatpipes need surface area to have the heat enter and leave them, and need a large enough temperature differential to evaporate and condense the water inside but also not too high or else the water will not condense.

[Enderman]

The things that CAN increase watercooling performance are

1) reducing thickness of the waterblock base, so the CPU is closer to the microfins

2) removing the CPU IHS for the same reason as above

3) removing silicon thickness on top of the CPU die, for the same reason as above, which intel actually did with their new CPUs.

4) Increasing conductivity by using a material such as graphene or diamond instead of copper, which is nearly impossible

5) Increasing the fin area by making the microfins taller/longer (although this would reduce the flow speed between them, and manufacturers like EK have probably found the sweet spot in their R&D)

[Sophia_Borjia]

A heat pipe water block could be made better and more expensive then a water block, not the same as tank around air cooler or water through heat pipes in air cooler.

Imagine something like this with thicker top with micro fins cut into it making direct silicon contact on one side and and in direct water contact on other face.  It would be very expensive, but if intel makes a 900w tdp hedt cpu/spaceheater someday, it could be worth the trouble.  

[Spindel]

I know I might be wrong in this, and there are many good (and also some not so good) comments. But I will power through with my idea (which is easy to do because I will never build it myself and be disproven if I’m wrong). Also because I’m stubborn.

Just the fact that water can carry away roughly 4000 times as much energy (per volume) makes this idea tingle in me.
 

While the above idea is null compared to a standard watercooler you also have to consider that a heat pipe moves an insane amount of energy utilizing that sweet sweet phase shift enthalpy.

 

Almost feel like I should order some parts home (If LMG doesn’t make a video of it first  ) and dabble with it against a basic resistive heater just for kicks  
 

 

[Windows9]

It might work, but at that point just buy a Chinese water block for $15

[Spindel]

Ping @AlexTheGreatish you need to settle this.

[Sophia_Borjia]

I considered heat pipe to water, went with 360mm radiator aio found on sale for $69 instead.

 

I was looking for all metal water block with microfins.  I had planned to rethread block to accept AN -4 fittings for ptfe/stainless hose to a liquid to liquid heat exchanger.

 

Reason for all metal waterblock is it would be used as part of custom heat pipe, acetone under partial vacuum in loop. Acetone will eat through most common plastics and o-rings.

 

Worked it out on paper, it could have cooled with no pump or fan, with a large external passive radiator, 5 gallon bucket for water tank, and would not function if case is put on side instead of upright(no wick in diy heat pipe). Figured size for wattage that would blow fuse at wall, that way no hardware configuration would be to much for the loop. 

 

Custom watercooling is easier and cheaper, when that is case it may be good time to consider a more normal thermal solution if possible.  If built problems not considered would probably show up in use, like the whole room water cooling LTT did.  My plan was about as crazy.

[AlexTheGreatish]

You can add water cooling to heat pipes, we did it both in the RED Camera Cooling and in the Water Cooled Network Switch, so you CAN, but in most situations you shouldn't.

 

Basically to remove heat you need to minimize the resistance to heat transfer.  Heat pipes are really good at having a very low thermal resistance, but they do have one.  In an air cooler this is good because it allows for the fins to be larger and more conveniently placed for good cooling, but in a water cooler you'd just be adding unnecessary resistance.

 

For what you're proposing we can think of it like a bunch of springs:

TIM = very stiff spring

Fins in water block = medium stiff spring

Heat pipes = basically a slinky

 

When added in series it is going to be the TIM and Fins that contribute the most to the overall stiffness of the spring, whereas the heat pipes "spring" immediately collapses and basically does nothing.

 

There is also the problem of heat pipe dry out.  This is when a heat pipe gets too hot for the liquid inside to return to the source of heat, and then it starts to have a high thermal resistance.  Water coolers work best when there is enough heat that air coolers are no longer effective (~400W).

 

TLDR: Heat pipes are good for moving heat from one place to another.  In a water cooler you can just dump all the heat directly into the water so no need to move it with heat pipes.