How not to design whole room water cooling

The way Linus is doing it.

I know he's only released part one of his video so far, but when the motivation behind a whole room water cooling solution is to remove heat from their office, it's not very promising. There are two main problems that I see so far.

  1. He mounted a radiator directly to the outside of their cases. This is when the water will be at its warmest, meaning that the highest heat transfer rate from the coolant to the surrounding will occur right at the computer case (heat transfer is directly proportional to delta T). There should be no radiator's at the computer to minimize loss of heat from the coolant inside his room.

  2. He has a sh*t load of bare copper piping running inside their office. That means he has a shit load of surface area between his coolant and a material with a very high thermal conductivity and a relatively high heat transfer coefficient to air.  i.e. that copper piping is acting like a brilliant heat sink, and all that heat is going right back into his room. 

All of that adds up to the overwhelming majority of the heat going back into their office. The concept is neat, and it could have been made to work, but the execution looks very poor so far. I suppose he could wrap the copper piping in insulation later on down the road, but my hopes aren't very high for someone who mounted a radiator right to his case when his goal was to get the heat out of the building.

1 Like

the problem with water cooling is that theres a limit due to the ambient temperature in your enclosed area

maybe adding a condenser or using a cold storage area and water proofing your components would be the solution as condensation would be a big problem

i do recall someone saying that water cooling your rig could result in even higher temperatures in a enclosed space so yea after my first exos disaster and using that stupid hd putty i've stuck with aircooling

then again some people make some nice displays with the tubing so yea still worth a shot if done properly

I agree with the things that you were saying. I think that it would have been a much better idea to use rubber hoses instead of copper pipes. That would insulate it until it got outside much better than copper. Also, they could have used a car radiator or two instead of a pack of pc radiators. I am unsure of how well it would work if they made it so that the water went raditor->firs pc -> second pc -> third pc ->..........-> radiator. The last few pcs in the line wouldn't be cooled very effectively. That might be why they mounted some radiators on the cases. What they could have done to solve this is have a second radiator in the middle of the loop to get the water cool again. Having a massive reservoir would be helpful as well.

You know what, now that I think about it, because the outside temperature is lower than the indoors temperature during the winter months in canada, it is very possiblie (i'd wager very likely in fact) that the coolant get's chilled below the ambient temperature in the computer case (it may even freeze on cold days!). If it is far enough below ambient, Linus might find that condensation is building in is computers. And yes, your room get's warmer when watercooling because when you remove heat from one component, it must be moved somewhere else (or converted into another form of energy, but that is not the case here). That heat goes to your room (or wherever your heat transfer surface is). Watercooling removes more heat from your components than aircooling, thus it puts more heat into your room. Linus could have avoided that if he insulated all his piping and didn't have a radiator mounted right to his case, then the majority of the heat would be trapped in the coolant until it reached the radiators outside.

I don't know what piping configuration they are using, but if it is piped in parallel then all PC's would be cooled similarly. And I agree, rubber/pvc piping would be preferable. 

Parallel would work better than a linear configuration. I am just concerned about the pressure that would be at the joints if the pump is pretty strong. I am guessing that they have it all covered what with his dad being a plumber and all.

What I would have done, personally, would be to have a car water pump->car radiator-> parallel to all pcs -> repeat. With rubber hoses instead of copper pipes. Cheaper and insulated. Don't know how well the joints would work with the hoses though. Soldering the copper is really strong. I think that a splitter between the hoses would work well enough.

1 Like

They could have probably used metal compression fittings to connect the rubber tubing. It may not be quite as strong as a soldered connection, but I would expect it to hold up to reasonable pressures. While the fittings would dissipate some heat, it will still be extremely minor compared to a room full of bare copper piping.

1 Like

I agree with all of this. But we havent seen all of the videos yet, so it is a possibility that the copper pipes will be covered with an insulation later on. I would go this rout for durability, and running rubber tubes up walls would be ugly and hard to do. 

With the amount of pipes shown on the walls, I would expect it to be parallel. I couldnt see them running it in series.

 

Also, and I could be wrong about this, but this is how I would do it, personally. The radiators on each of the PCs could actually be there to WARM the water closer to room temperature before entering the PC. They would be very good at stopping the condensation from happening. 

But like I said, this is all speculation, and how I would do it, personally.

It would go 

Outside cold Rad > Pump > Copper tube > Warmer Rad > PC > Insulated pipes > outside

Keeping the intake pipes exposed copper would also cool down the room significantly.

If they are expelling the heat into the room it isn't the wisest choice. They need a chiller unit that expels the heat  to the outside.

They could put in sacrificial zinc anodes in to fix the corrosion problem.

They said that installing AC or similar would be better, but they're doing this for the "wow" factor.

HAHAHA, click on my profile then explain to my why I am wrong. Then I will educate you.

If they used the rads on the cases to warm the water up, then they would need a condensation tray because any of the condensation that would form, would form on the warming rads. Could be that the rads that they showed were for separate loops should they need to break off from the big loop.

You can get a new car radiator for ~$100 where I am. That is not bad at all (though they might just have the extra rads lying around from reviews and whatnot). Cleaning out the rad with a filter before use is a good precaution, and I would expect that they should do that if using a car rad.

Literally nothing in his comment is even questionable. All of it is spot on. I am not sure what you are on about.

I would still be concerned about the coolant in the outside rad freezing. I don't know where exactly they are planning to put the rad, but I would mount it in a garage type space where the coolant hopefully won't be exposed to sub freezing temperatures (though I guess it may occasionally dip below freezing even in an attached garage) 

Did he ever say where the rads would be installed?

well i kinda like their idea, and as he said it is not really about reducing ambient, but on spot heat. also said that it is not really an effective way to make a room cooler, and they do it because they can. i look forward to see how it ends. it  was entertaining to watch for me.

What do you mean by "on spot heat?" My understanding was this project was started because LTT employees where complaining about the room (i.e. ambient) temperature being too hot. Instead of installing an AC, linus decided to move the heat produced from the computers outside. The only problem is it doesn't appear as though he is moving near as much of that heat outside as he could have.

i meant that guys's pc that was perma rendering or whatever, having high temperatures and stuff they even showed putting a thermometer on it. it's at the really beginning of the video. i am not too good with thermodynamics anyway. what my point is, let's wait how it ends and judge by the results :)

Well in the video the radiator they showed was being flushed. The pipes were being dry fitted so, there wouldn't be any insulation or lagging at this point. 

Heat would not get trapped as long as there is turbulent flow in the line. 

Water cooling vs air cooling will impart the same amount of energy to the room. Water will just reach equilibrium temp quicker. But none of this matters since the rads will be outside. But using a chiller or having it as part of the HVAC would be best since you can control the target temp easier. With the rads being out side you are stuck with the lowest temp being the outside ambient temp.

What I am interested in is what they plan on using as the pump hopefully they use a centrifugal pump so they don't over pressure the system and cause leaks. They will also need a surge volume to deal with thermal expansion. The surge volume can act as a source of static head (back pressure) for the pump suction.

Also since there is so much area in the system chemistry will be a concern. Corrosion can very easily eat through those small copper pipes. So daily ph test are a must so is having a chemical treatment and Ion exchanger is key (a house hold water treater will suffice).

They also need valves so they can deal with issues without having to re-purge in the entire system. Ball valve would work just fine since there is very little pressure drop across the valve. 

Source: 8 years of industrial experience.