Whole Room Water Cooling Part 4

[TheGrumpyDwarf]

gate valves will leak eventually. they restrict flow to some extent too.

full bore butterfly or lever valves are a far better option (these are ball valves).

 

valves on the one pipe system you describe will not provide a bypass, the system will just stop working if it is turned off.

 

Nomadluap is recomending an upgrade to the system to the one you describe. valves fitted on each drop on that system will work. unfortunately it is still a one pipe system & is inefficent. though it is used in some industrial/commercial installations.

[nekro]

that would only be better for the first PC in the loop, every pc there after would be getting hotter and hotter water.

i supposed the resistance of the waterblocks are about equal right? with the diagram the guy post a lot of the water flow  will get the less resistance way which is the central pipe. I do not knoiw the Vancouver's summer weather but i think it is higther than the in da house ampient temp so in the new office if they will use the same cooling again they need to put it in the winds direction I do not know how practical the watercooling project will be if it would  be merged with a geothermal heating/cooling. 

[LabRat]

gate valves will leak eventually. they restrict flow to some extent too.

full bore butterfly or lever valves are a far better option (these are ball valves).

 

valves on the one pipe system you describe will not provide a bypass, the system will just stop working if it is turned off.

 

Nomadluap is recomending an upgrade to the system to the one you describe. valves fitted on each drop on that system will work. unfortunately it is still a one pipe system & is inefficent. though it is used in some industrial/commercial installations.

you have to build it in.( the bypass ).

[Kevguy]

The loop needs some high point vents installed.

[TheGrumpyDwarf]

you have to build it in.( the bypass ).

then you end up with a load of tee's, just what you thought the problem was & still end up with a one pipe system.

one pipe systems have not been used in the UK in residential properties for perhaps 30 years.

 

& yes Kevguy aav's would help. i suggested it in my first post.

[Paranoid Kami]

Cover the end with a rag when you empty the water from it so it doesn't splash everywhere.

[Takata]

-snip-

 

I think because they chose to not insulate the piping (well part 5 isnt out yet, but it doesnt look like they plan to) I think the opposite effect will happen and the room will actually be WARMER, because the hot water inside the pipes can transfer their heat to the other side of the copper.. 

You're forgetting that the ability of the radiator and fans to transfer heat to the outside air FAR exceeds the ability of the copper pipes transfer heat into the room air. The room couldn't possibly be warmer than before, since before, ALL of the heat from the computers was getting transferred into the room air.

 

Maybe placing the "cold" line close to the floor and the "hot" line close to the ceiling would have made it easier to bleed the air out of the pipe network, since the main pump won't have to push air bubbles through pipes that goes _down_ 3 metres.

 

Good job troubleshooting that huge loop linus. ^^

 

@TheGrumpyDwarf: Care to show us a diagram of your proposal?

[CaptainDan]

Everyone knows what the true ultimate small adjustment tool is.

 

I still don't see why they didn't go with standard schedule 40 pvc pipe. Would have been so much cheaper and easier to work with, no soldering or anything. Just cut, fit, glue fit done.

 

Awww yeaaaah Top Gear  B)

 

IMO, the PVC just wouldn't look as good in this setup. If you were going to go with PVC, I'd almost prefer to hide the piping within the walls. More work, but the cleanest look for sure. 

 

Though I imagine they were also considering the ease of taking this stuff down in the future with the idea of a new office someday lol

 

Badass project either way.

[LabRat]

then you end up with a load of tee's, just what you thought the problem was & still end up with a one pipe system.

one pipe systems have not been used in the UK in residential properties for perhaps 30 years.

 

& yes Kevguy aav's would help. i suggested it in my first post.

 

 you don't understand the set up. yes, you will have tees. if I said where the pipe drops down you drop down with an ell ( not tee off ). add a small piece of pipe, a tee, a small piece of pipe in the other two ends of the tee each with a valve on it. then a small piece of pipe and a tee going up and down on the other side ( an H pattern if you will )... ... one to another valve that will isolate the pc ( as did the first valve ).. there will be 3 ball valves in close proximity. If I could draw and post a pic I would. but you seem to have a little knowlegde. you should be able to figure out what I'm saying.

 right now it's not properly looped.

 

 EDIT: this is heat loops I'm describing not fresh water lines.

 

 

guess I could add my old water set up here. back in the 478/939 days i took a "radiator" from an old beat down air conditioner. soldered fittings 1/4 O.D. CXF adapers on it and fitted that with male x barb adaptrers. got a cheap pump/heat sink from some store somewhare with 2 fittings and an old box fan. Ran some clear plastic tubing from the pet shop aisle in the local grocery store. filled the system with automobile coolant ( anti-freeze ) and ran it. did a smash up job for quite a while. fan only needed to run on low so the noise wasn't a problem. kinda crazy set up but worked better than Linus'.......... LOL

[Nomadluap]

that would only be better for the first PC in the loop, every pc there after would be getting hotter and hotter water.

The difference would be trivial for two reasons:

 

1. The temperature delta of the water traveling through a loop once isn't really that high

2. The main pump has power that is much higher than the individual pumps in the computers themselves, so that the water ejected by the first computer is mixed with water fresh from the radiators.

 

The temperature difference between the first and last computers shouldn't be more than a few degrees.

[little_shiet]

Man this looks gorgeous

[TheGrumpyDwarf]

You're forgetting that the ability of the radiator and fans to transfer heat to the outside air FAR exceeds the ability of the copper pipes transfer heat into the room air. The room couldn't possibly be warmer than before, since before, ALL of the heat from the computers was getting transferred into the room air.

 

Maybe placing the "cold" line close to the floor and the "hot" line close to the ceiling would have made it easier to bleed the air out of the pipe network, since the main pump won't have to push air bubbles through pipes that goes _down_ 3 metres.

 

Good job troubleshooting that huge loop linus. ^^

 

@TheGrumpyDwarf: Care to show us a diagram of your proposal?

ok i have tried my best to provide a diagram. just a rough idea you understand. now let's see if this works.

 

i think i may have missunderstood you LabRat.

 

p.s. i can't spell. but i try my best

 

[GorbazTheDragon]

OK, so watching this last video pretty much confirmed some of the disappointing  problems with this setup. I think in the future such a system deserves far more planning than it got.

 

First off, the pipe positioning. Other than making it easier to put through walls, there is really no reason the main hot-cold pipes (the copper ones) should be run at ceiling height. As was seen the air was almost impossible to force down the pipe due to buoyancy and the smaller pumps being unable to push air. Running the main pipes along the ground would have significantly reduced the head pressure needed to get the water through the loop in the first place and made disconnecting/reconnecting computers in the later phase much easier, as the computers could be connected through upwards pointing T junctions which would naturally contain the water inside them.

 

Secondly, I have a feeling that the pipes on the primary runs were way too narrow, especially the 1/2' OD tubing connecting the primary pump to the reservoir. I am unsure of how much this affected the system, but it doesn't seem well thought through.

 

Valves on the T junctions for connecting the PCs would have been a really easy way to save a lot of time bleeding the loop. You could simply open the valve slightly to let the air out, and close it immediately when the water pushes through. Additionally, it could save a lot of mess when trying to connect new computers to the loop, especially if used with the ceiling height pipe runs.

 

Another flow restriction related problem would be the outside reservoir. Flow wise it would have been better to connect these in parallel, although this might cause some issues with certain rads pulling through more water than others. The expensive solution would be to have each radiator in series with a small pump like the PCs have, however this need could potentially be bypassed by using more restrictive tube diameters after each T junction with the radiator.

 

You will also need to be careful that the outside tubes do not receive too much sunlight, I am 100% certain that they will be damaged by continuous exposure.

 

Connecting the computers in series would not be good due to the high restriction of flow through the waterblocks. Additionally it would make it a lot harder to install new systems and integrate them to the loop.

 

Running the hot primary line along the ceiling might be problematic with the head pressure, as that will mean the main pump will need to first push the water through the waterblocks, and then up 2.5m of pipe. As it is right now, the height added by the initial climb seems to somewhat neutralize the pressure required for the waterblocks and part of the climb back up to the top of the loop.

[NumLock21]

If you guys decide to re-do the watercooling at the new office, the design should definitely be changed as follows:

 

What? All that hard work and now move to the new office? I say put some wheels on it and drive the damn house to the new location!

 

Bubbles? I like bubbles!

[Nomadluap]

In a closed system like this, the height of the tubes only really matters when bleeding the air and filling the pipes for the first time. Once the loop is fully filled, the total gravitational potential energy of the system stays constant and changes in elevation have no effect on pressure. The extra head pressure required to push the water up to the supply line will be counter-balanced by the suctioning effect of the water falling down the other side.

[Guest Kloaked]

I don't think I've rolled my eyes in one sitting as much as I did reading some of the comments here and on Youtube.

 

LMG, thank you for doing something cool like this. I like it when you do things like this, and I'm glad sponsors dig it as well - enough to provide equipment to do so. 

[bohica]

Since you already have the system partially in place there are a few things you can do to mitigate problems, make things a bit more constant, and simpler to add on other pc`s in the future. The biggest question I have is are the pipes providing cooling that are the farthest away from the pump in use? If not, they will hold air in the system forever... :wacko:

It should be similar to bleeding the brakes on a car, go furthest away from the pump and work your way towards it. Any lines not in use should be closest to the pump and capped off, and valves for flow control would help to optimize performance.

[LabRat]

now that we all have gotten our comments in.......... time to grab the plumbers favorite tool, the sawsall. Cut out the old and build us another one.

[Psykomantis00]

This whole thing was poorly planned and implemented. It was a bad idea to start with but for some reason it was green lit. 

[Takata]

ok i have tried my best to provide a diagram. just a rough idea you understand. now let's see if this works.

 

i think i may have missunderstood you LabRat.

 

p.s. i can't spell. but i try my best

 

I... don't really get this. ?

I think a break tank is kinda like a reservoir, but you seem to have one off a branch immediately downstream of your pump, instead of upstream. ?

Dunno how you can be sure that the pump will always have water on its input (esp. while filling the loop) when the first things upstream of it are a long pipe and the PC waterblocks.

Auto air vent? Dunno what that is, but I'm guessining it will allow air to escape the loop while you're filling it.

 

 

I think this series could use a Q/A video kinda like the mineral oil series.

[PorkFish]

Love this project, but had a hard time understanding, what problem you were having with the unused pipes. Hope to understand more, once you give us the complete rundown.

 

EDIT: OK, after rewatching twice and reading the comments in this diskussion, I finally do get the problem you were having. I guess my english isn't as good as I thought it was 

[TheGrumpyDwarf]

I... don't really get this. ?

I think a break tank is kinda like a reservoir, but you seem to have one off a branch immediately downstream of your pump, instead of upstream. ?

Dunno how you can be sure that the pump will always have water on its input (esp. while filling the loop) when the first things upstream of it are a long pipe and the PC waterblocks.

Auto air vent? Dunno what that is, but I'm guessining it will allow air to escape the loop while you're filling it.

 

 

I think this series could use a Q/A video kinda like the mineral oil series.

the break tank is a device used industrial aplications. not really necessary here.

it would have been better if i had just drawn in a double check valve or non return valve if you like (so as a liquid can pass in under pressure but not come out. when necessary a stirrup pump for example could be attached in order to introduce fluid.

it makes no difference where in the system this is connected.

the system would be filled without any pump running. that would be asking for trouble. with the valves open to one pc at a time. the liquid would be under pressure, air will naturally gravitate to the highest point & be forced out of the air vents.

the auto air vents work a little like the non return valve. it will let air out but not liquid.

once the system is filled & under pressure (0.3 bar perhaps) & vented, if balanced i doubt there would be any need for individual pumps within each pc. though that is debatable.

[acturisme]

As a pipe fitter this project fascinates me. There are some good suggestions here. Isolation valves would have gone a long way to fixing the bubble issue by running them out of the system one loop at a time. I'd have used quarter turn ball valves because they offer the least restriction with easiest installation and use. Also they show direction of flow when properly installed. Globe valves (the one with garden faucet valve handles) would be a poor choice as these add resistance to flow. I'd also have put the res at the top of the system. That would allow the bubbles to escape into the res without having to travel down. Also seriously, get some proper tools. Use line wrenches to tighten line fittings. STOP using pliers. Makes me cringe every time. But, you guys did great otherwise. My applause.

[LeafInTea]

My guesses for this project is that it will heat up the room rather than to cool it down.

[BenHTBDT]

These videos had extremely good production value, but I don't understand why they took 4 months to be released... The build was done by August, and we get videos by December.