Where is my water going?

Joined
Oct 23, 2018
Messages
1,011
One of my loops is very slowly losing fluid. The fluid in the reservoir has been steadily dropping by 2-3mm per day for the last two weeks. I use UV reactive fluid specifically so I can spot leaks more easily, and yet I can't find anything when I shine my light all over. Yesterday, I went through and loosened and retightened every single connection. Today, the level is down again.

WTF is going on?

I even put a piece of masking tape on the res yesterday to mark the starting level and verify once and for all that this isn't just in my head.
 
How recently did you fill it?

Sometimes as you run the loop over time more trapped air makes its way out of the loop and the level gradually goes down until all the air is all out.

Even once all the air is out the level will continue dropping due to evaporation. Even with a sealed loop some evaporation occurs straight through the tubing, but it shouldn't happen this fast.

Make sure you don't have any leaks in difficult to spot locations.
 
Last edited:
How recently did you fill it?

Sometimes as you run the loop over time more trapped air makes its way out of the loop and the level gradualyg goes down until all the air is out.

Even once all the air is out the level will continue dropping due to evaporation. Even with a sealed loop some evaporation occured straight through the tubing, but it shouldn't happen this fast.

Make sure you don't have any leaks in difficult to spot locations.
this
 
How recently did you fill it?

Sometimes as you run the loop over time more trapped air makes its way out of the loop and the level gradualyg goes down until all the air is out.

Even once all the air is out the level will continue dropping due to evaporation. Even with a sealed loop some evaporation occured straight through the tubing, but it shouldn't happen this fast.

Make sure you don't have any leaks in difficult to spot locations.

The level was pretty stable for about two months and then I noticed it dropping a little over two weeks ago. Most recent fill was yesterday after doing the purge and checking every single fitting. One of the nice things about having a gajillion QDCs in the loop is that I can pre-fill it really nicely on my bench before putting each section back into the machine and re-connecting it. The drop from yesterday might be a big air bubble that I missed, so I guess I'll keep an eye on it over the next few days just to make sure.

I've checked for leaks about a dozen times with no luck. I even went with UV fluid and kept all ports and connections facing the same direction so that I could track down leaks more easily. The best laid plans of mice and men....


What sort of tubing?

Alphacool Alphatube 13/10 soft tubing (PVC).
 
The level was pretty stable for about two months and then I noticed it dropping a little over two weeks ago. Most recent fill was yesterday after doing the purge and checking every single fitting. One of the nice things about having a gajillion QDCs in the loop is that I can pre-fill it really nicely on my bench before putting each section back into the machine and re-connecting it. The drop from yesterday might be a big air bubble that I missed, so I guess I'll keep an eye on it over the next few days just to make sure.

I've checked for leaks about a dozen times with no luck. I even went with UV fluid and kept all ports and connections facing the same direction so that I could track down leaks more easily. The best laid plans of mice and men....




Alphacool Alphatube 13/10 soft tubing (PVC).


Hmm. If you don't mind my asking, where are you located?

Did the relative humidity in the air by any chance drop (or the outside temp get much colder) during the affected time period?

It could be evaporation, but it does seem awfully quick.
 
Last edited:
Just be careful. You don't want your pump to run dry as this could be harmful.

But I agree, you want to be just another as agressive as you can be with your case gymnastics to get the air out. Much more so than seems reasonable.
Yeah, great advice, I do not run the same configs around here that everyone else does.

If you have the pump/and/or/res mounted up already, don't run your pump dry. And if you are hard-tubed, this is likely not a good solution for you. But even if you are, tilt it up on is corner and burp it as well you can. Both directions.

If he is not seeing leakage, like there is a lot of air somewhere that is getting disturbed.
 
One question: do you run your pump(s?) at reduced speed for silence?

I do, but for the first several weeks after a fresh fill I crank it to 100%.

In the past I've filled a loop with the pump at about 60% speed, and run it for weeks. The fluid level had stopped dropping, I couldn't see any bubbles or hear any gurgles - I was sure I was completely bled.

One day I was playing with my settings and ramped the pump up to max. Immediately, air started belching back into the reservoir. Those bubbles had been in my loop the whole time and I had no idea.

My build is an awkward beast to tilt as well, so I decided that 100% pumps after a fill was mandatory from then on out.
 
I'll give the 100% pumps a shot and see what happens. Would just a day be enough?

Dropped another 3mm in the last 24hrs.
 
I'll give the 100% pumps a shot and see what happens. Would just a day be enough?

Dropped another 3mm in the last 24hrs.
I run mine full speed for a couple weeks after a fill, but I only run my computer a few hours a day too. I'd think a solid 24hrs should do it.
 
I run mine full speed for a couple weeks after a fill, but I only run my computer a few hours a day too. I'd think a solid 24hrs should do it.


Every system is different, but with mine, a full speed pump is no louder than slower settings, only yje pitxh is different, so I just run mine at 100% all the time.

I get the 2-3mm drop OP posts about, but we are talking every couple of months, not daily.
 
After I put my recent updated loop in I found a really crappy bubble in the side of the block. I questioned the acrylic siding up front but thanks to UV liquid and a UV flash light a rather large bubble stood out once I got in there looking for anything I could find for higher temps. Was going to pull the rad next for some altered angles but after reseating the block it fixed that temp concern.

Then as others have mentioned, living in Dallas its been rather dry vs the humid summer when I put the loop together. Mixed with me pushing my OC's a lot more than ever before as the loop has cleared some other small air I noticed my level dropping a bit. I was surprised there was so much air left in the nooks and crannies but it seems to have stabilized.

I also poked my head in and checked the tube ends as they go over the nipples and saw a few that while not leaking were reacting under UV light so I wiggled a paper towel in to suck up the fluid and pushed the tubing back on (after an hour of gaming so the tubes would be soft) and tightened the bolts on better.

I think its done bleeding and I check the tubing every few days in case the hot to cold cycling pulls them off a bit again but thus far it looks clean. I also prefer a decently aggressive ramp up on the fans and pump when temps get to a certain point which helped clear some bubbles. It can take some time but the general use causing the flow rate to pick up and slow down can work some bubbles free. Not all, but some.
 
Last edited:
if your loop has more than one rad it may take alot longer to get the air out of them. ive found that with 2-3 rads its just about impossible to get all of the air out of the loop during the initial purge. mostly because they are in much different locations to one another. ie- its easy to get the air out of the front rad but really hard to get the air out of the top or bottom rads. and sometimes getting the air out of one rad while tipping and revolving. you may accidentally get air going back into another rad. one thing to try, if your rads have vent screws, try venting them after the loop has become heat soaked. if air comes out repeat that every couple of days until nothing but coolant comes out. lastly be sure to slide or roll your case over and check for any wet spots underneath your case. most likely you just have air in the rads and itll take a while to get it out.
 
if your loop has more than one rad it may take alot longer to get the air out of them. ive found that with 2-3 rads its just about impossible to get all of the air out of the loop during the initial purge. mostly because they are in much different locations to one another. ie- its easy to get the air out of the front rad but really hard to get the air out of the top or bottom rads. and sometimes getting the air out of one rad while tipping and revolving. you may accidentally get air going back into another rad. one thing to try, if your rads have vent screws, try venting them after the loop has become heat soaked. if air comes out repeat that every couple of days until nothing but coolant comes out. lastly be sure to slide or roll your case over and check for any wet spots underneath your case. most likely you just have air in the rads and itll take a while to get it out.

I have two rads in mine. After filling it I do a series of case gymnastics, tilting it side to side (all the way flat on its side) and back to front as far as I can with it running. I also start and stop the pump several times. I also shake it back and forth a bit. I do this repeatedly until a good chunk of time has passed without hearing any rushing air bubble sounds.

I never get all the air out, but it is good enough at this point. The rest of the air generally seems to work its way out over the next few weeks or a couple of months. I continue to top it off every now and then. I'd say I never lose more than a half an inch to an inch or so after doing this.

Also, it is always a good idea to keep a little air in the top of your reservoir. Don't fill it all the way to the top. Air is compressible, so it helps with temperature changes. Water is not very compressible, so if loop volume or water volume changes with temp, there is a greater chance the heat expansion/contraction will do damage to the loop if there is no compressible air that can absorb some of those changes. It looks better when it is completely full, but to be on the safe side.
 
It's probably hard to tell because of the spaghetti, but I'm running one rad per loop with three loops in total. I [finally] posted build details here.

Ran the pumps at 100% since Tuesday and just turned them back down to check the levels. Both reservoirs are down about an inch. A full inch drop on these reservoirs is like 50% of the volume in each waterblock. I can't fathom how there would be a bubble that big in the blocks. Starting to look more certainly like it's not bubbles at this point (IMO). I refilled the reservoirs and marked the levels, so I guess I'll do one last level test with the pumps maxed for a couple days and see what happens.

I double checked the rad vent caps and all four bleed fluid when I loosen them. Also did another check with my UV flashlight and came up empty.

I'm confused.
 
It's probably hard to tell because of the spaghetti, but I'm running one rad per loop with three loops in total. I [finally] posted build details here.

Ran the pumps at 100% since Tuesday and just turned them back down to check the levels. Both reservoirs are down about an inch. A full inch drop on these reservoirs is like 50% of the volume in each waterblock. I can't fathom how there would be a bubble that big in the blocks. Starting to look more certainly like it's not bubbles at this point (IMO). I refilled the reservoirs and marked the levels, so I guess I'll do one last level test with the pumps maxed for a couple days and see what happens.

I double checked the rad vent caps and all four bleed fluid when I loosen them. Also did another check with my UV flashlight and came up empty.

I'm confused.

Wow, how come you decided to that route? You'd have a much more effective loop with all three blocks in the same loop with one pump. Temps would be lower.
 
Wow, how come you decided to that route? You'd have a much more effective loop with all three blocks in the same loop with one pump. Temps would be lower.

One loop is dedicated to the CPU, and the GPUs are in pairs with one pair on each loop. I did this for redundancy and serviceability.

The water level in the CPU loop seems good. The two GPU loops are the ones which are running into the issue with water level.
 
The water level in the CPU loop seems good. The two GPU loops are the ones which are running into the issue with water level.

Hmm. This is odd. The quantities you are losing seem too large to be explained by air in the system considering each loop is so small. I'm thinking maybe there is a slight drip somewhere that is evaporating before it is pooling up and looking like a detectable leak.

I would run it and go in there with a strong flashlight and look around each fitting and see if there is any moisture at all.

One loop is dedicated to the CPU, and the GPUs are in pairs with one pair on each loop. I did this for redundancy and serviceability.

Ah. Generally you wind up with lower temps if everything is in the same loop. CPU and GPU(s) are rarely at full load at the same time, so by sharing the radiators across the entire system, your typical temps will be much lower. You also only have one pump dumping heat into the loop, instead of multiple pumps.

I do see how it is easier to replace each component if you have them dedicated, but you could also use quick disconnect fittings to accomplish that.
 
It's probably hard to tell because of the spaghetti, but I'm running one rad per loop with three loops in total. I [finally] posted build details here.

Ran the pumps at 100% since Tuesday and just turned them back down to check the levels. Both reservoirs are down about an inch. A full inch drop on these reservoirs is like 50% of the volume in each waterblock. I can't fathom how there would be a bubble that big in the blocks. Starting to look more certainly like it's not bubbles at this point (IMO). I refilled the reservoirs and marked the levels, so I guess I'll do one last level test with the pumps maxed for a couple days and see what happens.

I double checked the rad vent caps and all four bleed fluid when I loosen them. Also did another check with my UV flashlight and came up empty.

I'm confused.
Are those quick disconnect fittings? Could the leak be internal to those, such that the coolant evaporates and leaves residue inside it where you can't see?

Maybe crack one open and see if there's coolant residue crust in there.
 
Ah. Generally you wind up with lower temps if everything is in the same loop. CPU and GPU(s) are rarely at full load at the same time, so by sharing the radiators across the entire system, your typical temps will be much lower. You also only have one pump dumping heat into the loop, instead of multiple pumps.

I do see how it is easier to replace each component if you have them dedicated, but you could also use quick disconnect fittings to accomplish that.

My usage generally has either all GPUs idling or all GPUs at full load. Currently running stock clocks and 123% TDP on the cards, and they peak at ~50C depending on the workload with 22C ambient air. 480x60mm of radiator is still a lot for only 700W of load.
 
The filler for my loop is a small length of tube with a marble to seal it.
This makes it easy to see any difference in water level.

When I fit new tube the water level drops a lot for the first month or two and at the same time creates a vacuum (the tube squashes until I release the marble).
I top it up every 3 days ish to prevent tube kinks.
My assumption is the tubing takes on water for a while until it is saturated.
After a few months only a small top up (a few ml) is required every few weeks, the water is replaced by air so it no longer creates a vacuum.
ie water permeates out the plastic (evaporates) and at the same time air permeates in.

This could explain some of the water level drops you see.
 
Back
Top