Outrage: cooling 10 comps with an Uber pump and a bong

[Balinju]

Sorry forgot attachment

[hara]

Thanks for your math redleader! I appreciate your calculations. That means that about 70liters would have to be used everyday. WhAT happens to the calcium deposits afterwords (and shit)?

[jaydee]

Originally posted by redleader
Math time!

10 comps and a "550w pump" probably works out to less then 1kw IME. Don't believe the pump power ratings, typical is generally half that, if as much. (Like my 24w Eheim 1250 that uses 9w). Lets say 1000w.

Enthalpy of evaporation is 22kj/mol for water at 101.3kpa and (IIRC) 20C. That works out to one mol lost every 22 seconds. A mole of water is 18g and a liter has a mass of 1kg, so you'd loose a liter every 1222 seconds or about every 20.5 minutes. That means 70.7 liters (about 18 gal) evaporated per day.

Plus the evaporated water gets replaced with cold water from the tap, so you get some "free" cooling right there. Lets use the above figure for simplicity of 70.7 l. Thats 70700g at say 10C below bong temp so you'd get almost 50 minutes of free cooling. Of course if you want sub ambient temps you'd loose that much trying to cool incoming water . . .

Not too bad compared to AC.

Edit: Just checked and according to my dad a cubic foot of water (~7.5gal) is 10 cents or so here in Arizona. If thats correct, then you'd be looking at just 24 cents a day assuming you can run the coolant at outside airtemp. Thats way less then a tenth of what AC would cost. I think I see why swamp coolers are so popular here.

You may want to check my math though.

Yeah that's not good. Replacing that much water is going to be a bitch to keep the right amount anti-growth/corrosion stuff in the system. Running tap water through the system is a bad thing. You will have to hire a guy at minimum wage to stand there and dump mucho dollars of watter wetter a day if you want the system to last anyway. Water cooling basics, tap water no good! And you STILL need A/C. Back to basics, you are moving heat from one place to another place. The heat that the bong is evaporating has to go somewhere. If it is in the room it will go in the room and you will need a/c to cool the room or it will heat up.

If the bong is outside then you need to cool the air going into the bong as if it is 120F outside on a summer day you will be adding heat to the sytem instead of taking it away. Read what has been said first. I wouldn't want my computers running well over 120F.

There is no way that somesort of AC is not going to be used. And furthermore the computer room itself is STILL going to need A/C!!! CPU's are not the only thing putting out heat. With a lot of computers on one semi sealed room it will heat up fairly fast. All the componets put off heat. Even the water blocks themselfs will put of some heat. And you also have to add in all the other costs! Running those high speed fans in the bong is going to suck up a lot of power.

[hara]

Then what? Car rad with the integrated fan and al blocks? Of course, we would specify that some items eg: psu will need to be watercooled and in some cases scsi hds. The room then won't need an ac, just an extractor.

[jaydee]

Originally posted by hara
Then what? Car rad with the integrated fan and al blocks? Of course, we would specify that some items eg: psu will need to be watercooled and in some cases scsi hds. The room then won't need an ac, just an extractor.

Power supplies, North Bridges, Ram, Video Cards (if they are attached anyway), network cards, hard drives,anything that generates heat. Thats just the computers. Then you need to figure out what other equipment in the room that generates heat.

If this is a possible design for a server room then you need to think alot more professional than a car rad and AL blocks! Anything you use for a heat exchanger, be it s radiator or bong or whatever, will need to have cooled/heated air going through it to maintain a level of consistancy. You cannot cool/heat air without active cooling/heating. Active cooling being AC of somekind active heating a heater of somekind.

If this is for someone's business you need to find out all the variables of where this equipment will be installed. You will need to know how the piping will be routed, the amount of drop and rise the fluid will be traveling, the size of the piping, ect ect ect... It all starts with where the stuff will be installed. Then you take what you know there and start designing the system around that. Once you know how to route the piping and where all the equipment is going to go, then you need to calculate the heat dissapation of all the equipment that will be cooled. Once you find the total wattage of heat output of all the equipment then you can send that information to a company that designs heat exchangers. They will then make a custom exchanger to handle the load and they should give you the needed flow rate variables. Then from that you will have to figure out how to get that amount of flow through all of the attached water blocks on all of the equipment and through the exchanger where ever it will be installed. They should also give you an idea of how much active cooling/heating you will need to cool/heat the air that is going through the exchanger per ambient temp to maintain a safe level of operating temperature forthe equipment.

For example X is ambient air temp Y is the temperature that the air will need to be maintained for a consistant operatating temp, Z is the amount of cooling/heating needed to maintain Y.

If X is 100F then they will tell you that Z amount of cooling will be required to keep Y consistant. OR if X is -30F they will tell you how much HEATING will be required to keep Y consistant!

In something as expensive and important as a server room you want your operating temps to be within a certain range and consistancy. That is the only way to properly control things IMO. I would always want the temps of the cooled components to be consistant within a set limit. To cold and it can cause problems and to hot and it can cause problems. The heat exhanger will have to be retro fitted with some kind of electronic ciruitry that will turn on the AC or the heater to maintain the Liquids temperature to keep the components at the safe operating temperature. And of course this would all have to be approved by certified engineers in plumbing, electrical, and heating/cooling and then approved by the local building code people.

This is not going to be acheived by a car rad and aluminum blocks.

Now if you are just doing this in your house for a personal project then a car rad and AL blocks would probably be and option. But if you took this to a person in charge of a expensive sever room that is critical to be up 24/7 you will end up in the hospital getting the security gaurds boot from you a$$. Or at least you would in the USA.

So maybe explain in greater detail what it is you are thinking of doing with this project. Is it for a personal thing or someones business? And whatever the answer to that question is you then need to be a lot more spacific about the equipment being used (ALL of the equipment) and how it will be installed and in what enviroment (what kind of room, top floor/bottom floor, size of room ect. ect..).

Otherwise we are just wasting our time here.

[hara]

I'm baffled. If it were for a business we wouldn't be here anyway. It is a school project, a fierce competition which is based on originality and skill. We came here to learn as we're only 17 and have little experience with this field.

I have agreed with my friend that a radiator is the way to go. Anyone have a different opinion, please speak now.

Some components eg: ram, NBs, video cards will not be watercooled. Basically, what will happen next is to arm a computer with hardware, build the blocks, and watercool. We don't have 5 pcs so we'll just watercool one and then place a heater element to simulate the load of other computers. It is basically a scaled model. We are cheap bastards after all and will not afford more than Lm150 ($400) between us 3. Most of the things will be either:

a) stolen
b) lended
c) salvaged

We would then have to buy some materials eg: copper stock. The next thing to do is determine the material of the heat exchanger that my friend found at his dad's work place. It is quite large, bigger than a truck rad.

Thanks for your replies and keep them comming. They motivate us

[hara]

Using a bong would have been more original. Is there some hope left?

[jaydee]

Originally posted by hara
I'm baffled. If it were for a business we wouldn't be here anyway. It is a school project, a fierce competition which is based on originality and skill. We came here to learn as we're only 17 and have little experience with this field.

I have agreed with my friend that a radiator is the way to go. Anyone have a different opinion, please speak now.

Some components eg: ram, NBs, video cards will not be watercooled. Basically, what will happen next is to arm a computer with hardware, build the blocks, and watercool. We don't have 5 pcs so we'll just watercool one and then place a heater element to simulate the load of other computers. It is basically a scaled model. We are cheap bastards after all and will not afford more than Lm150 ($400) between us 3. Most of the

Still need more info. A school projects, but what are the goals? And what are the "virtual" variables. In other words what would this stuff be doing and what would you want it to do in reality?

So far what I got from it is you want to cool 10 computers with liquid cooling but you only have 1 computer to work with so it will be on a 1/10 scale? If so this is a good in way as it will be a lot easier to construct. The thing to remember though is you still need to think of it on a real time 1/1 scale. This I think would still rule out the bong theory.

If you are just going to use 1 computer completly throw out all given information in this thread for what to use! You should have pointed this out from the start. You can just use a normal heater core for the rad. You will not need a mondo car rad if it is just for 1 computer. Pretty much just build a standard water cooling system and then when you do the presentation times the numbers by 10!

For example if that heater core is 10x10, then times that by 10. If you had 10 computers you would need a rad 100x100 to acheive the same results as a single computer with the 10x10 rad! Then do that with all the numbers. If you have X amount of airflow on the 10x10 rad, then for a 100x100 rad you would need X times 10.

You are obviously not going to get real time results as adding more computers is a lot more involved than just multipliying the numbers, but if all you have is one computer to work with then thats about all you can really do.

[hara]

I have already made my own personal system

http://www.geocities.com/neville_grech/wickedpc.html

but there's nothing original with that. We are going to SIMULATE the heat load of a number of computers. The goals are to construct something which is syentific, original, and exclusive. Who wins will go abroad to show off the project elsewhere. It's the originality that's lacking. We would have to also show some numbers and stuff

[hara]

The pump will be operational because it was blocked and f-ed up really hard (ask balinju if you don't believe me).

Balinju, urihom x'kella l-pompa, turihomx zobbok imma... bil-mod.

[jaydee]

Originally posted by hara
I have already made my own personal system

http://www.geocities.com/neville_grech/wickedpc.html

but there's nothing original with that. We are going to SIMULATE the heat load of a number of computers. The goals are to construct something which is syentific, original, and exclusive. Who wins will go abroad to show off the project elsewhere. It's the originality that's lacking. We would have to also show some numbers and stuff

Ok if you are going to simulate the heat load then don't bother with a computer at all. What you need to figure out though is what exactly IS the heat load going to be? Or is it just going to be a made up number not based on any spacific equipment? Is that where the 2000watts came from?

So lets break down what needs to happen.

1) Figure out the heat load.
2) Find a accurate and consistant way to simulate the load.
3) Construct a block that will absorb the load and have the liquid going through it. (should be a part of #2 just aswell make it all one peice)
4) Figure out what the heat exchanger is going to be.
5) Pump (already done)
6) Tubing
7) Temperature and flow monitoring.

Now what would be cool would be to add a system that could adjust the water temperature to maintain a spacific temperature range. Something like this would be a MUST on a real system. For example if you have the heat exchanger on the roof of the building and gets 120F outside you would need to cool the air down in order to keep the water in the certain temp range. BUT you would also need the ability the HEAT the water. Say again the exchanger is on the roof and a snow storm blows in and it drops to 20F outside. You sure the hell would not want the water to drop below whatever the temp is in the computer room (obviously the computer room will be considerably warmer in this senario) because that would cause massive condensation on the installed blocks and piping through the bulding and in the computer room possibly causing sever damage! So you also need the ability to add heat to the water to maintain the certain temp range.

This is probably out of your range though.

[Balinju]

Ok I found the pump after about 5 months of rest (rest for the pump not for me )

I took it to my workroom, so first to open this shit of a pump, i had to cut all the screws because they were all rusted and no spanner, socket or glipper in the world could untie them. (Only god knows how much penertating oil I used, but all in vain)

After I managed to open it, it was full of sh*t (in the sense of, salt[from the water], and many other things like calcium ...)

So I cleaned the first part, but when I opened the middle part and took out the shaft, please do not try to immagine in what condition the ball bearing was!!

After a lot of hammering, I managed to make the shaft turn, but turn inside the ball bearing not with the ball bearing.

Ok, this took me more than 2 hours, not a joke.

Then, at this point I decided to take out the ball bearing from the shaft to clean the shaft and replace the ball bearing. This was not a hard time because as I touched the ball bearing with the hammer, it "dissolved" in pieces smaller than 1mm squared
. I know it is a little bit fanastic, but please believe me, it is true.

Now try to immage the rest of the pump. When I will be ready with it, I will tell you if the work was don for nothing or not. I could not finish the work because I could not go and but parts

Thank you for your help and comments.


jaydee116, we cannot do what you said, (use a small car rad and multiply by 10).

This is a project and has to be original, that is why we decided to use a bong at first. It is true that we will set the stuff for only one computer but we will
simulate the heat formed by other computers with a heating element, like the one used in a kettle.

Is this explanation better??

[hara]

My question is this, why add heat? the cooler the better

[jaydee]

Originally posted by hara
My question is this, why add heat? the cooler the better

I told you why! Cooler is not better in something like this. If the liquid is allowed to get colder than the temperature that is in the computer room then condensation (WATER) will form on the water blocks and piping causing water to get all over the computers!!!!!

It is the same principle as pelt/vapochill cooling. That is why you have to insulate your components to protect then from the condensation that they produce.

You guys have to ask yourselfs what is more important. Originality or something that actually works!?

[nicozeg]

Originally posted by jaydee116
I told you why! Cooler is not better in something like this. If the liquid is allowed to get colder than the temperature that is in the computer room then condensation (WATER) will form on the water blocks and piping causing water to get all over the computers!!!!!

It is the same principle as pelt/vapochill cooling. That is why you have to insulate your components to protect then from the condensation that they produce.

You guys have to ask yourselfs what is more important. Originality or something that actually works!?

You're complicating more than necessary; The systen's already making a lot of heat. If ambient is too cold, you only need to stop airflow thru rad. Generaly, coolant temp 10ºC colder than room dont cause condensation.

[jaydee]

Originally posted by nicozeg
You're complicating more than necessary; The systen's already making a lot of heat. If ambient is too cold, you only need to stop airflow thru rad. Generaly, coolant temp 10ºC colder than room dont cause condensation.

A lot of heat yes, enough to keep the liquid room temp on a cold winter day? absolutly not! Even if the airflow is shut of. I have had to work on the roofs of building in a middle of a snow storm at 0F on a windy day. There is plenty of cold to drop the temps down below the room temp. All it takes is ONE extreamly cold windy day to destroy thousands upon thousand's of $'s of equipment! Would YOU take that risk? I know I wouldn't.

There are ways around this though without adding a heater. You can make a bypass valve that will direct the warmed water back into the main line without going through the exchanger, but you DO NOT want to completly shut off the liquid flow through the exchanger as that will freeze it and wren the bypass valve closes you will have no flow. Then that would be a good test for the overheat sensors you would have to attach to each peice of equipment that are connected to a main shutoff switch.
There are reasons liquid doesn't cool this amount of high $ equipment already. If it was as easy as some make it out to be in this thread it would have already been done a long time ago!

The military does this but not on such a large scale, and they always have a backup system in case of disaster. This is not possible/economical for a server room application.

[hara]

This would be a problem, if our country was that cold

Seriously, the lowest one sees in winter is about 6C at night, and not all the years will winter be colder than 12C. We could put a sensor to shut off the airflow. That would definately work.

[jaydee]

Originally posted by hara
This would be a problem, if our country was that cold

Seriously, the lowest one sees in winter is about 6C at night, and not all the years will winter be colder than 12C. We could put a sensor to shut off the airflow. That would definately work.

It doesn't matter what the average temperature is. All it takes is one abnormally cold day to ruin everything. Even if it was the first day abnormally cold day in last 3000 years, that is all it takes. Are you willing to risk the cost of all that equipment? I know I woldn't be.

Anyway we can put this issue aside for now and concentrate on the project at hand.

Will post some more after work.

[jaydee]

Originally posted by Balinju

simulate the heat formed by other computers with a heating element, like the one used in a kettle.

Is this explanation better??

Explanation is still unlear.

Anyway you say you are going to use a heating element? Do you plan on just 1 2000watt element? I don't know if that would be a good idea. It would probably be better to have multiple heat sources spread out to mimic a real situation. In a server room you are not going to have 2000watts all in one spot, unless it is on fire . I am sure the liquid would react considerably different at a 2000watt load as oposed to multiple 100watt loads spread out. But this will not be an option in your situataion. Would it be possible to make 4 500watt loads?

And next how do you plan on installing the (these) heat load(s). Are you going to put them in a big copper block and let the copper radiate the heat? If this is possible then you can also make the cooling liquid run through that very block. That would be less work. The element has to be enclosed somehow in order to contain the heat. Then what are you going to use to monitor the wattage output? Numbers are going to be useless unless you know exactly how much wattage is being distributed. Might just be able to get away with a thermal probe instead and just use temperature as the variable instead of wattage.

[hara]

It has been decided!

We have a space constraint of 1x1x2 meters. We'll use bongs either in this arrangement (viewed from above)

..\...../
...OO
....O
.....|
Where the circles represent the bongs and the lines represent the "squirrel" cage blowers. IGNORE THE DOTS.

Yet another idea:

..OOOO
......|

Same as above, one squirrel cage blower distributed among 4 bongs. In both cases, the bongs will be glued with each other.

The pump should be functional. It is just a working model, it only has to work for the amount of time it'll be exibited

Before attaching the bong(s) to a computer, we'll first simulate the heatload and record the temperature of the water accordingly. After that, one can easily calculate the maximum number of computers that can be attached to the system. Finding a heater element is an easy task but finding a variable resistor that can handle 10 amps at 240V is a different story.

Anyone care to enlighten us?

[hara]

Correction: The pump is 370Watts. One has to remember that it's the concept which counts.

[bigben2k]

Varying the power supplied to the fan may not work: it'll depend on the type of motor.

It would be far easier to restrict the airflow, unless you just can't live with the noise.

[hara]

You didn't understand. It is to vary the heat generated by the heating element.

[bigben2k]

Sorry, you cought me: I wasn't paying attention!

That would be very easy actually, except for the large power rating. there are many ways to reduce the power from an outlet. In this case, clipping the AC wave would work just fine, as long as the power is used only by the element. Power diodes.

[hara]

It's the high power rating that's worriying me :shrug:

Can the power then be calculated using an ammeter?