Dismiss Notice
Join Physics Forums Today!
The friendliest, high quality science and math community on the planet! Everyone who loves science is here!

Water cooling

  1. Sep 14, 2012 #1

    I am trying to design a water cooling system that could cool down my computer from 100C to about 40C, with 600W of energy being provided by the components. Could a system that moves 0.2l of water at 0.1kg/s from a water tank to the computer components to a radiator and from there back to water tank and around again, cool the system?

  2. jcsd
  3. Sep 14, 2012 #2


    User Avatar
    2017 Award

    Staff: Mentor

    How do you get 0.1kg/s with just 0.2kg of total water? A round-trip time of 2 seconds with a water tank looks strange. In addition, with a pipe cross-section of ~1cm^2, this would require a water speed of 1m, which is "a bit" more than I would expect.

    Anyway, with your numbers and assuming a perfect radiator, the initial water temperature is similar to the room temperature or ~20°C. To heat 0.1kg of water to 40°C, you need 100g*4.2J/(g*K)*20K = 8.4kJ, while your computer just produces 600J per second. The heat capacity of water is more than enough, if you get that flow.
  4. Sep 14, 2012 #3

    I'm hoping to get the water to move at roughly 0.1kg/s by using this pump:


    It's rated for 400 litres per hour (max), so a bit over 0.1 kg/s.

    Do you think I should have more water with the water moving at that speed? What do you mean by a perfect radiator? The radiator that I would use would use forced convection of air, with about 80CFM being blown on it from 4 sides. Is there a way to get a ballpark estimate about how big the radiator would have to be in order to dissipate the heat gathered by the water into the air?
  5. Sep 14, 2012 #4


    User Avatar

    Staff: Mentor

    You're mixing systems of units, but..

    CFM*DT*1.08 = BTU

    Lbw is pounds of water
    DT is delta-T.

    So lets say you have a 70 F room and want your water to max out at 95F or 35C. Using the equations I posted gets you a required airflow of 75CFM. So you're probably ok there. A good heat exchanger will get the water down to about 75F. I'll let you calculate the required flow rate of the water...

    Let me ask you this though: how do you know the computer uses 600W? Is that the rating of the power supply? Odds are it is using much less than the rating.
  6. Sep 15, 2012 #5
    This sounds like a complicated problem since you have transient variables. I have some expertise with a thermal simulation program I might be able to help you. PM me if you are interested. I use it for work almost every day and we work with hydronic heating and cooling systems.
  7. Sep 15, 2012 #6
    Re:comment on Water cooling

    i read ur problem, and i think u should consider some points
    1)computer system can't alone rediate all the heat u must gather all the heat for the purpose of ease
    'cause if u r going make water flow in all that delicate computer system it might turn into mess

    2)i rather suggest u to use another coolent instead of water, which should not be fire catcher and nor could cause the short ckt in ckt if leaked

    3) u should use the some nellpolish remover type coolent, 'cause in water cooled engines water is used 'cause it doesn't allow temp to icrease beyond 100 deg C.
    In that order nellpolish remover type thing can easily gain input heat at 20 to 30 deg C from the system and can also rediate to the heat sink easily and ur overall system temp will be less than desired easily

    4) Apply large alluminium heat sink at system and make coolent flow through it using gravity it will make cooling system requirements very less and another heat sink at radiator u r also going need a exhaust fan outthere. which will cool down the coolent to be again liquid.

    5) Use "Calculus of Variation" to determine the coolent amount

    and here u are done
    u can reply me on purn.shodh@gmail.com
  8. Sep 15, 2012 #7


    User Avatar
    2017 Award

    Staff: Mentor

    Perfect as in "cools water to environment temperature" - unrealistic of course, but as the calculation shows there is a big safety margin of a factor 10 (or more, if your actual power is lower).
Share this great discussion with others via Reddit, Google+, Twitter, or Facebook