Pumpless (convection based) chilled liquid cooling idea

I saw a post on another board where a member made a convection based water cooling set up. Basically, he connected both tubes from his CPU waterblock to a radiator. It did, in fact, work. The radiator removed heat from the coolant and this coolant sank, while the CPU heated coolant that rose up to be cooled. He has water flow through the system without a single pump.

I was wondering if this concept could be taken a few steps further...

What I have in mind is a copper block with ports in the top and bottom. Let's say you wanted to cool the CPU and two GPU's. There would then be six ports. The hot coolant would enter at the top and exit from the bottom ports. Mated to this block would be the evaporator from a phase change unit. The copper block and evaporator would be mounted high in the case, around the PSU so the hot coolant can rise. The reason for this idea is to eliminate pump noise and heat, make the cooling system more compact and cool multiple components to near phase change temps without having multiple evaporators and/or P.C. units.

This idea is probably hard to visualize, and I am working on a 3D rendering, but are there any thoughts? Am I crazy?

I dont see your temps being that low with this kind of setup.

Epox4life said:

I dont see your temps being that great with this kind of setup.

Click to expand...

I doubt it would be any worse than a chilled liquid cooling system with a pump. You would effectively have heat pipes between the waterblocks and the evaporator.

MarkS said:

I doubt it would be any worse than a chilled liquid cooling system with a pump. You would effectively have heat pipes between the waterblocks and the evaporator.

Click to expand...

I bet your temps would be a lot worse than a system with a pump. What you are describing would have to have a very small evap to fit compactly inside your case and your system would then not have any capacity. The advantage of a chiller is being able to use a large res with a huge evap to keep the liquid at -40c. No way a small evap keeps all the liquid in your loop at that temp.

Ad Rock said:

I bet your temps would be a lot worse than a system with a pump. What you are describing would have to have a very small evap to fit compactly inside your case and your system would then not have any capacity. The advantage of a chiller is being able to use a large res with a huge evap to keep the liquid at -40c. No way a small evap keeps all the liquid in your loop at that temp.

Click to expand...

OK, this point has me confused. Don't phase change systems use a very small evaporator? The evaporators I've seen can all fit in my hand.... With room to spare...

Dealing with different amount of heatload for one. And heatremoval as well as design of it. It need to be tuned for at least 300w for all the loops you have and units such as Mach II GT and vapochill LS runs out of steam once it hits 150w mark.

units such as Mach II GT and vapochill LS runs out of steam once it hits 150w mark.

Click to expand...

Mach II and Vapochill LS can handle 200watts past that they start not cooling so well.

MarkS said:

OK, this point has me confused. Don't phase change systems use a very small evaporator? The evaporators I've seen can all fit in my hand.... With room to spare...

Click to expand...

Yes they use a very small evaporator but on a very small surface area. Its like trying to cool a watercooling loop with a radiator the size of a 40mm fan. You have to take into account the thermodynamics of liquid when designing a chiller.

MarkS said:

OK, this point has me confused. Don't phase change systems use a very small evaporator? The evaporators I've seen can all fit in my hand.... With room to spare...

Click to expand...

Direct Die Phase Change units - when you have the phases changing directly on one spot (the cpu or gpu die) - have a small evaporator similar to the size of a heatsink.

Phase Change Water Chillers - when you use phase change to cool water to a very low temperature - requires a much larger evaporator often in the form of a curled copper pipe. This requires a larger evaporator size due to reasons such as:

- Direct Die directly cools the cpu. The phase change occurs in a space about 4 square inches. The Water Chiller cools the water, which in turn cools the components.

- As water passes over the chiller's evaporator, the heat from the water is added to the gas in the evaporator, thus making the water cold. As water continues to circulate, more surface area is needed in order to continuously cool the water.

Hope this clears up at least some confusion.

MarkS said:

I saw a post on another board where a member made a convection based water cooling set up. Basically, he connected both tubes from his CPU waterblock to a radiator. It did, in fact, work. The radiator removed heat from the coolant and this coolant sank, while the CPU heated coolant that rose up to be cooled. He has water flow through the system without a single pump.

I was wondering if this concept could be taken a few steps further...

What I have in mind is a copper block with ports in the top and bottom. Let's say you wanted to cool the CPU and two GPU's. There would then be six ports. The hot coolant would enter at the top and exit from the bottom ports. Mated to this block would be the evaporator from a phase change unit. The copper block and evaporator would be mounted high in the case, around the PSU so the hot coolant can rise. The reason for this idea is to eliminate pump noise and heat, make the cooling system more compact and cool multiple components to near phase change temps without having multiple evaporators and/or P.C. units.

This idea is probably hard to visualize, and I am working on a 3D rendering, but are there any thoughts? Am I crazy?

Click to expand...

This wont really work.

For one the Delta T has to be large enough, and it wont be.

Then coolant gets thick in chillers and does not flow. Try to use the same method with oil as a coolant, the heat will pool, and there wont really be movement. And at the same time unlese you are running something like methanol alcohol your coolant temperatures at the evap will drop so low the coolant will slush or frezze firther hindering or stopping the flow.

The flow if any would be so low that the heat from the CPU/GPU will not be removed fast enough, the waterblocks will heat up, and your computer will either burn or shut down. Your loop at the evap will be frozen and the blocks will be hot as hell.

A pump is a very small price to pay for superb preformance. Loosing the pump would do what? Save you 40w (if even that much...), thats $0.006 an hour on your power bill, and thats at crazy CA prices of $0.15 per KWh. Noise? not really if you buy the right pump. Space? lol... I dont see any point to this.

This is a thermosiphon...

MarkS said:

I saw a post on another board where a member made a convection based water cooling set up. Basically, he connected both tubes from his CPU waterblock to a radiator. It did, in fact, work. The radiator removed heat from the coolant and this coolant sank, while the CPU heated coolant that rose up to be cooled. He has water flow through the system without a single pump.

I was wondering if this concept could be taken a few steps further...

What I have in mind is a copper block with ports in the top and bottom. Let's say you wanted to cool the CPU and two GPU's. There would then be six ports. The hot coolant would enter at the top and exit from the bottom ports. Mated to this block would be the evaporator from a phase change unit. The copper block and evaporator would be mounted high in the case, around the PSU so the hot coolant can rise. The reason for this idea is to eliminate pump noise and heat, make the cooling system more compact and cool multiple components to near phase change temps without having multiple evaporators and/or P.C. units.

This idea is probably hard to visualize, and I am working on a 3D rendering, but are there any thoughts? Am I crazy?

Click to expand...

http://en.wikipedia.org/wiki/Thermosiphon
This technique has been used for the cooling of automobiles quite successfully. Now that I've found this thread, I'm going to read through it. Because right now, I'm in a situation where not having a pump would be nice. Or at the very least, minimizing the size of the pump.

Shingoshi

Holy thread bump!

Yeah! I know I'm guilty...

NoL said:

Holy thread bump!

Click to expand...

But sometimes those who have already been down the same road, can give the best directions. And I'm certainly looking for good directions.

Shingoshi

Read the thread, idea was shot down, doesn't work. Convection is incredibly slow.

I wrote my response after reading the first post...

NoL said:

Read the thread, idea was shot down, doesn't work. Convection is incredibly slow.

Click to expand...

I saw what happened to the idea afterwards. But I will go back and read it again. Something that should also be remembered, is that convection rate is a function of the liquids used. There are better liquids for convection than water. Ethanol is one of those. There are a few others, but I'll pass on mentioning them, since few would feel safe using them.

Another thing to notice, is the placement of the radiator is critical in the design of a thermosiphon. There can't be any place for liquid to collect. The heat source absolutely MUST be below the radiator. And there must be a straight vertical flow between the radiator and the cpu. See the link I gave above for Thermosiphon on Wikipedia. It will explain how critical placement really is.

Thermosiphons are strictly gravity based systems. The flow of the system is determined by the change in density of the working liquid. Simply, warm liquids are less dense than cool liquids. So warm rises, and cool falls. The rate of movement in the liquid is determined by the gross weight of the liquid in the reservoir compared to the weight of the liquid in the return line (going back to the reservoir). Since the reservoir's liquid will have a substantial mass, as that liquid cools it will present more weight on the warmer liquid, causing it to rise. So circulation won't be a problem, as long as you follow simple rules. And the larger the reservoir (in this case the radiator which serves that purpose), the more efficient the cooling of the system will be. That's why you need really large radiators for this to work. And it should also be noted, the larger the radiator, the higher the rate of circulation. So concerns about convection can be diminished accordingly. You're essentially doing nothing more than creating a very large heat pipe. Which is why ethanol is a better choice for a system like this. It evaporates more quickly than water, having a higher convection rate.

http://en.wikipedia.org/wiki/Thermosiphon

Shingoshi

EDIT: And using good fans will increase the rate at which the liquid cools in the radiator, causing it to become more dense more quickly. Again, that increases the rate of convection.