Do High Coolant Temps cause Leakage?

Discussion in 'Water Cooling' started by Zarathustra[H], Aug 26, 2018.

  1. Zarathustra[H]

    Zarathustra[H] Official Forum Curmudgeon

    Messages:
    25,237
    Joined:
    Oct 29, 2000
    Hey all,

    I popped open my case today to clean out some dust. When I did I noticed some evidence of dried coolant on the back plate on my GPU (right below the CPU block), and on top of my PSU (right below the GPU fittings). The spots were completely dry, but shiny and colored from the coloring and additives in the coolant, so whenever it leaked it was not very recent.

    I wiped up the dried coolant, and put paper towels underneath, and have been testing under full load all day today, with not as much as a drop coming out of those fittings. I simply cannot reproduce the leak.

    Then it struck me. I remembered this one night I started a gaming session, and accidentally left my fans off. It wasn't until quite a while of poor framerates I finally realized it had been throttling due to overheating. The coolant temp levels probably went well in excess of what is safe for the pump. Luckily the pump does not seem to have sustained damage.

    This has me wondering. Could the thermal expansion effects of heat have caused my fittings to leak during that episode, and now that the coolant temps are normal again (usually about 35C under high load when the room is not airconditioned at about 85F) the leaks are gone?

    If not, I can't explain it. Appreciate any thoughts.
     
    Last edited: Aug 26, 2018
  2. KazeoHin

    KazeoHin [H]ardness Supreme

    Messages:
    7,441
    Joined:
    Sep 7, 2011
    So, Tubing becomes REALLY SOFT at high temperatures. So much so that a tube that is ROCK SOLID at room temp, we're talking "Lift the entire PC by using this tube as a handle" solid, can easily pop out with slight tension when at 60+ degrees (Celsius).

    Not saying that is what's going on now, but it certainly could be.
     
  3. Zarathustra[H]

    Zarathustra[H] Official Forum Curmudgeon

    Messages:
    25,237
    Joined:
    Oct 29, 2000
    Sounds at least plausible then. I heated up the tube, and while it was hot, the seals around the fittings had a slow leak, but now that they are back down at sane temperatures, I can't reproduce the leak.
     
  4. KazeoHin

    KazeoHin [H]ardness Supreme

    Messages:
    7,441
    Joined:
    Sep 7, 2011
    I know it sounds whacky, but it's possible the pump-speed may be the culprit as well. High tubing flexibility + High fluid pressure may be the mix perfect for a slight leak. If may be that you've adjusted the pump speed downward after the leak happened, but before you noticed it, OR its a PWM pump that adjusts it's speed automatically thus it is REALLY hard to recreate the exact conditions to reproduce the leak, OR OR it's an incredibly slow leak, causing maybe a drop or two after HOURS of extended heat.

    I've seen some situations like this, check to see if you have any particular tubes at tight bends near or above the scene of the crime. I found out the hard way that some tube bends that are fit as a fiddle when cold can EASILY pop out when warmed up, and that 90 degree fittings are your friends in that situation.
     
  5. pendragon1

    pendragon1 [H]ardForum Junkie

    Messages:
    11,396
    Joined:
    Oct 7, 2000
    high temps cause the fluid to expand so if there is nowhere for it to go, like the res is full and sealed, then it could push past O-rings. it could also be the tubing getting soft/deformed as mentioned ^.
     
    cyberguyz likes this.
  6. cyberguyz

    cyberguyz Limp Gawd

    Messages:
    469
    Joined:
    Aug 28, 2014
    This. Particularly if you are rocking rigid tubing. Flexxy tubing has a bit of give when it warms up that rigid tube doesn't. It is always smarter to leave just a bit of an air bubble in your reservoir so that expansion/contraction compresses air rather than your fittings.
     
  7. Tsumi

    Tsumi [H]ardForum Junkie

    Messages:
    12,545
    Joined:
    Mar 18, 2010
    Probably a combination of all of this. High pressure due to high temperature caused leakage around soft tubing. Pressure lowered after cooling down, causing a vaccuum. If the reservoir hasn't been opened, it's possible that vaccuum still exists and is preventing things from leaking when warmed up.
     
  8. VanGoghComplex

    VanGoghComplex [H]ard|Gawd

    Messages:
    1,252
    Joined:
    Apr 5, 2016
    After filling and bleeding, I will typically thermally punish my system with the reservoir uncapped, getting the coolant temp 5°c over where it usually runs, and then capping the system while it's hot. This means that once it cools down I'm running under a slight vacuum, but that is preferable to the alternative!
     
    Zarathustra[H] likes this.
  9. VanGoghComplex

    VanGoghComplex [H]ard|Gawd

    Messages:
    1,252
    Joined:
    Apr 5, 2016
    I'd also argue that air trapped in your loop will expand far more with temperature than water will, so my guess is that a large quantity of air in your res will contribute to the pressure problem more than alleviate it.
     
  10. Zarathustra[H]

    Zarathustra[H] Official Forum Curmudgeon

    Messages:
    25,237
    Joined:
    Oct 29, 2000
    Well, to me the solution is to not be an idiot and forget the fans off again.

    When I first built the system I hooked up loop temp sensors to a fan controller and used it to control fan speed.

    The problem with this is that it would just soon up the fans on hot days, and down on cold days which was annoying.

    Using the motherboard CPU fan header wasn't very good either, as that didn't set fan speed based on CPU load only, not based on GPU load, and, if ifdle for a long time it would let the coolant get hot, such that when the CPU loaded up, the fans couödnt cool the coolant down enough to keep the CPU cool, leasing to a temp spike before reaching a stable temp at load.

    In the end I just gave up on any assemblance on automated control and just soldered my own manual dial PWM controller board and manually set the fan speed.

    Fans at ~480rpm are perfect for idle and light desktop use. Heavy games on a warm day can require fans at 1,250rpm and sometimes higher to keep the GPU at 40C or below.

    It geberally works pretry well, espexialy since I have a large volune loop with large res, so the loop temp changes slowly, and I have plenty of time to adjust if I need to. The problem is that this one time I just forgot. Oops.

    I should put a beeping alarm on there that goes off if the loop temp exceeds like 40-45C. That way I won't pull a stupid again :p
     
  11. VanGoghComplex

    VanGoghComplex [H]ard|Gawd

    Messages:
    1,252
    Joined:
    Apr 5, 2016
    I believe that you should buy an Aquaero.

    Of course, I believe that everyone with a watercooled system should buy an Aquaero, so take that for what it's worth. XD
     
  12. toast0

    toast0 Gawd

    Messages:
    825
    Joined:
    Jan 26, 2010
    Beeping alarm should do it -- if you're lucky, your BIOS has a temperature alarm, and you just need to configure it.
     
  13. Zarathustra[H]

    Zarathustra[H] Official Forum Curmudgeon

    Messages:
    25,237
    Joined:
    Oct 29, 2000

    Sadly, my motherboard does not have a connector for temperature probes.

    I could go by CPU temp, but that may not catch it if the major heat source is the GPU, and the CPU has low load.

    I did buy this little dedicated thermal control board which might do the trick though.
     
  14. Zarathustra[H]

    Zarathustra[H] Official Forum Curmudgeon

    Messages:
    25,237
    Joined:
    Oct 29, 2000
    I've always been interested in the Aquaero, but the last time I researched it there were so many different modules and configurations, most of them only on webpages in German (my German is very rusty) that I got confused and delayed shopping for one.

    That, and I also remember them being excessively expensive.
     
  15. VanGoghComplex

    VanGoghComplex [H]ard|Gawd

    Messages:
    1,252
    Joined:
    Apr 5, 2016
    They're pricey, but most of the price is the screen, the regular versions all come with one.

    http://www.performance-pcs.com/aquacomputer-aquaero-6-lt-usb-fan-controller-26462.html

    Dis right here is the one you want. No frills, just awesome standalone configurable cooling goodness.

    PM me if you have any questions - I absolutely love mine. It'll treat you right.
     
    Zarathustra[H] likes this.
  16. Zarathustra[H]

    Zarathustra[H] Official Forum Curmudgeon

    Messages:
    25,237
    Joined:
    Oct 29, 2000

    Hmm.

    But this one is just a board?

    I thought they were 5.25" bay mounted with screens?
     
  17. VanGoghComplex

    VanGoghComplex [H]ard|Gawd

    Messages:
    1,252
    Joined:
    Apr 5, 2016
    The "flagship" is 5.25" with a screen, yes, but frankly, the power of these units lies in the software. I don't see a purpose to the screen other than eye candy. After using the software I couldn't imagine the depth of configuration possible being available on a little LCD with a few buttons.

    The board is nice because you can pop four holes and mount it wherever is convenient for your cable runs. You don't have to see it - you don't even have to run the software anymore once you've set it up, because once it's configured the thing runs standalone, executing your configuration by itself.
     
  18. Zarathustra[H]

    Zarathustra[H] Official Forum Curmudgeon

    Messages:
    25,237
    Joined:
    Oct 29, 2000

    Hmm.

    The software appears to be windows only.

    If I configure the unit using the software, does it need to be running, or if I reboot into Linux would the settings remain and still be active?
     
  19. VanGoghComplex

    VanGoghComplex [H]ard|Gawd

    Messages:
    1,252
    Joined:
    Apr 5, 2016
    You don't need to be running the software for the unit to work. Your configuration is stored in memory on the board, so it can run completely independent of your OS.

    I can't say for sure if the board will have access to embedded temperature sensors in your hardware if it's running without the software. Personally that's a non-issue, because the only temperatures my Aquaero control on are the ambient air and the coolant temperature (or more specifically, the delta between those two), which it gathers via dedicated sensors plugged into the board itself.

    The only link my unit has to the rest of the system is a fake "tach" signal that it generates for the motherboard's CPU_FAN header. That tach signal will halt if my pump stops working or if the coolant temperature gets too high. Along with the tach stopping, an audible signal sounds from the unit: all this stuff was configured by me in the software.
     
    Chapeau likes this.
  20. IdentityCrisis

    IdentityCrisis [H]Lite

    Messages:
    104
    Joined:
    Oct 29, 2012
    I would think seepage around fittings would be pressure related? I don't see how high temps could cause that, but stranger things have happened.
     
  21. Zarathustra[H]

    Zarathustra[H] Official Forum Curmudgeon

    Messages:
    25,237
    Joined:
    Oct 29, 2000

    Well, 2+ weeks later at normal temps, and I have not been able to reproduce any dripping.

    There are multiple ways tempersture could cause leaking.

    1.) Heat causes the tubing to soften and softer tubes are more likely to give.

    2.) Thermal expansion and contraction. As materials heat up they expand. Different materials expand at different rates. Generally metals expand more than other materials. So, if the loop gets really hot, the metals in the fittings could expand just a tiny bit more than the rubber seals/tubing resulting in a leak.

    I'm leaning towards the latter being the case in my system, as it has stopped since cooling down. This would be expected behavior for thermal expansion/contraction.

    If the soft tubing were the cause, I'd imagine the leak would be permanent. The tube could loosen in the fitting while soft, and not go back again when it hardened up.
     
    Last edited: Sep 10, 2018
  22. IdentityCrisis

    IdentityCrisis [H]Lite

    Messages:
    104
    Joined:
    Oct 29, 2012
    in the case of an extremely high increase in temperature as you said you experienced, I can see it, are these compression fittings? I've no experience with them, I've always used standard barbs with either a slightly smaller ID tube than the barb or a zip tie wrapped around it, never experienced a leak. I wonder if maybe the fitting needs a slight tightening? like a quarter turn or something?

    Were you able to determine exactly where it was leaking from on the fittings?
     
  23. Zarathustra[H]

    Zarathustra[H] Official Forum Curmudgeon

    Messages:
    25,237
    Joined:
    Oct 29, 2000
    Nope. I never actually saw the leak. Just dried coolant on the GPU backplate (from one of the CPU fittings) and on the top of the PSU (from one of the GPU fittings).

    No matter what I have done since, I have been unable to recreate the leak. If I can't recreate the leak, I can't tell exactly where it is coming from.
     
  24. VanGoghComplex

    VanGoghComplex [H]ard|Gawd

    Messages:
    1,252
    Joined:
    Apr 5, 2016
    Have you tried gaming for a few hours with your fans off? XD
     
  25. IdentityCrisis

    IdentityCrisis [H]Lite

    Messages:
    104
    Joined:
    Oct 29, 2012
    That's a bummer. I'd be interested to see if it was that vacuum theory mentioned above, have you opened your res up then did some benchmarking to get it warm? I suppose it a good thing you cant recreate it, but man that would make me nervous.
     
  26. Zarathustra[H]

    Zarathustra[H] Official Forum Curmudgeon

    Messages:
    25,237
    Joined:
    Oct 29, 2000

    Lol. I'm trying to avoid doing that again :p
     
    toast0 and pendragon1 like this.
  27. IdentityCrisis

    IdentityCrisis [H]Lite

    Messages:
    104
    Joined:
    Oct 29, 2012
    LOL I wanted to suggest this, but I would feel terrible if that caused another leak with more than just some spilled coolant as a result, haha. But it is one way to find out.