Graudally Decreasing Flow

[hihihi8]

Hi guys, so it's been almost two weeks since completing my first watercooling build. It is a very restrictive loop consisting of 2 universal GPU blocks, 1 CPU block, 2 ram blocks, a 480mm Rad, and a lot of sharp 90 degree bends. It is all powered by a D5 Vario running on speed 5, which gave 0.74GPM which I considered acceptable. (Adding another pump is not an option)

 

Now the interesting part:

First few days of the first week, flow rate from my aquaero sat at around between 0.73 and 0.75 GPM (169LPH or 2.8LPM)

However, over the the second week the flow rate started to slightly decrease, first starting to read around 0.71-0.72 GPM, and now sits stable at 0.70GPM and hasn't budged for hours (min 0.70, max 0.70) I haven't changed a single setting, and the pump is stable at 4830RPM the whole time with little fluctuation (it's connected directly to the +12V Rail)

 

What could be causing this? I have the aquaero's alarm set to report blockage/pump failure at less than 0.6GPM, tho it likely won't hit that low. So far I have a few hypothesis:

 

- Unbled loop: the loop was a pain in the A to bleed, and I gave up 20 minutes in because I was afraid of wrecking the case since this thing weighs like 35KG-ish. There is still a section of the tubing that stays empty after i shut off the pump, but it mysteriously goes away after I turn the system on and cannot figure out where the air is stuck.

 

- Radiator Debris in Filter: I did clean the radiator with hot water, but after reading some posts on how Alphacool radiators are notorious for being very dirty out of the box, I'm reconsidering the possibility that the RAD hasn't been cleaned out properly.

 

Any ideas? thanks!

 

PS: that section of tubing that fills with air I mentioned earlier, is the highest section of the entire build, and is a long horizontal tube (hard tubing), so it is very difficult to bleed properly.

[NoobCase]

If you have the option of leaving the fill port in the res unscrewed do so and let the pc run for a few days. Should get the air out. You can help it along by shaking the case a bit now that you don't want to tilt it. Turning the pump speed up and down (if you can) can also help release trapped air.

Make sure to top of the res when done. The less air in the ress the better.

[hihihi8]

4 minutes ago, NoobCase said:

If you have the option of leaving the fill port in the res unscrewed do so and let the pc run for a few days. Should get the air out. You can help it along by shaking the case a bit now that you don't want to tilt it. Turning the pump speed up and down (if you can) can also help release trapped air.

Make sure to top of the res when done. The less air in the ress the better.

Trying that right now! thanks!

I'm also gonna try putting a large thermal load on the loop in the hopes that the temp change will squeeze some bubbles out. (No need to worry about water spewing out the top, as my loop has a secured and dedicated Y-split for filling and compensation for heat expansion)

[NoobCase]

4 minutes ago, hihihi8 said:

Trying that right now! thanks!

I'm also gonna try putting a large thermal load on the loop in the hopes that the temp change will squeeze some bubbles out. (No need to worry about water spewing out the top, as my loop has a secured and dedicated Y-split for filling and compensation for heat expansion)

NP - keep me posted.

[hihihi8]

On 10/20/2016 at 11:01 PM, NoobCase said:

NP - keep me posted.

Hmm... It's been a few more days, and flow has now decreased to ~0.67-0.68GPM, and this trend is starting to make me worry... as always, pump speed is at 4830rpm stable, but flow keeps on decreasing. If this has to do with the radiator being dirty and clogging up the filter partially (thank god not the water blocks, would be a nightmare to clean those), then I might as well just leave it be until 6 months later when my next coolant change is scheduled. If it drops below 0.60 before then, I guess my only option is to take the loop apart and clean out the filter, rad, and flow meter thoroughly.

[airdeano]

if that tube is still not able to hold coolant when shut down, then you still have an air pocket. this is not allowing a hydraulic base.

the coolant is pressurized to move through the tubing, but once it hits the air pocket, the air compresses and halts the flow rate.

 

have a picture of the layout and tubing runs?

[hihihi8]

2 minutes ago, airdeano said:

have a picture of the layout and tubing runs?

yeah sure, I'll try to upload them in a sec

[hihihi8]

 

[img=http://i.imgur.com/W7WO0Pj.png]

[img=http://i.imgur.com/xoUmzmS.jpg]

[img=http://i.imgur.com/xb1Eqo9.jpg]

[img=http://i.imgur.com/fCvwDV0.jpg]

[img=http://i.imgur.com/AFE2paE.jpg]

As you can see, it's quite a big air bubble, and I have not found a way to fill it since it is the highest section of the loop. Partially my fault for designing it that way  

 

[gamerking]

18 minutes ago, hihihi8 said:

 

[img=http://i.imgur.com/W7WO0Pj.png]

[img=http://i.imgur.com/xoUmzmS.jpg]

[img=http://i.imgur.com/xb1Eqo9.jpg]

[img=http://i.imgur.com/fCvwDV0.jpg]

[img=http://i.imgur.com/AFE2paE.jpg]

As you can see, it's quite a big air bubble, and I have not found a way to fill it since it is the highest section of the loop. Partially my fault for designing it that way  

 

<blockquote class="imgur-embed-pub" lang="en" data-id="W7WO0Pj"><a href="//imgur.com/W7WO0Pj">View post on imgur.com</a></blockquote><script async src="//s.imgur.com/min/embed.js" charset="utf-8"></script>

 

you need to  have 6 in's of straight tubing no bends on both ends of the flow meter  if not that will screw with the flow reading alot . for the aquacomputer flow meter . which one mps 200,400,100 ? if you can lay it down on its back so is flat and can run the pump for awhile you should be able to get alot of the air into the res . you will have to tip it back up a few times to fill it up with more coolant . this is the simplest way i see of bleeding it , without redoing some of the bends

[hihihi8]

2 minutes ago, gamerking said:

you do have 6 in's of straight tubing no bends on both ends of the flow meter right

thanks for pointing that out, totally forgot about it! Unfortunately I don't think thats possible in my case, I'll try and see what I can get, but I might just end up having the tubing taking a big detour inside the case which isn't really what I was hoping for. 

5 minutes ago, gamerking said:

for the aquacomputer flow meter

I'm using the Koolance INS-FM17 in this case, with the cable modded to go with the aquaero

[gamerking]

Just now, hihihi8 said:

thanks for pointing that out, totally forgot about it! Unfortunately I don't think thats possible in my case, I'll try and see what I can get, but I might just end up having the tubing taking a big detour inside the case which isn't really what I was hoping for. 

I'm using the Koolance INS-FM17 in this case, with the cable modded to go with the aquaero

ok i dont know for koolance one they may not require the 6ins .   i think it needs it from a few pics off of google i found .  you can just male to male 2-4in one  off the back of the rad then very slight bend on soft tubing you can get away with for the output of the flow meter . like a 2 foot section for that so there is litle to  no turbulence caused . the front where you have the ram going to the rad that would be a perfect place

[hihihi8]

Ok so I was overcome by my OCD, and took the loop apart. I found quite a bit of "gunk" like metal scraps and tiny paint chips in the filter, so I guess if must've been the radiator being dirty. Aside from that, gamerking pointed out how turns before a flowmeter would cause incorrect readings.

6 hours ago, gamerking said:

if not that will screw with the flow reading alot

Is this because of turbulence? BTW I am interested in knowing how it would affect readings, as so far readings have been stable besides the gradual decreases. I can hypothesize how turbulence would cause lower-than-true readings due to how the water is churning and causing areas of different pressure, but thats just my educated guess and I don't have any support. Can you or anyone confirm this?

 

Anyways, I don't think I'l be going through the fuss to change tubing routes since I am more than happy with current temperatures, and the primary purpose of the flow meter is for me to monitor if anything bad happens (and so the aquaero knows to emergency shutdown in the case of something).

[gamerking]

15 hours ago, hihihi8 said:

Ok so I was overcome by my OCD, and took the loop apart. I found quite a bit of "gunk" like metal scraps and tiny paint chips in the filter, so I guess if must've been the radiator being dirty. Aside from that, gamerking pointed out how turns before a flowmeter would cause incorrect readings.

Is this because of turbulence? BTW I am interested in knowing how it would affect readings, as so far readings have been stable besides the gradual decreases. I can hypothesize how turbulence would cause lower-than-true readings due to how the water is churning and causing areas of different pressure, but thats just my educated guess and I don't have any support. Can you or anyone confirm this?

 

Anyways, I don't think I'l be going through the fuss to change tubing routes since I am more than happy with current temperatures, and the primary purpose of the flow meter is for me to monitor if anything bad happens (and so the aquaero knows to emergency shutdown in the case of something).

im just going by what i got from aquacomputer from on their site buying the meter and emailing them goes either way can show its lower or higher then its really the flow rate .but their meters are a bit more picky on the setup i think  .

 

  there was testing done years ago that showed it caused issues the bends/90's too close to the flow meter but i cant seem to find it. the only thing google spits out is a site that hasn't been paid for in a year + . there is some charts floating around still from it but not the article

[hihihi8]

Ok, so I've pinpointed the cause of restriction to one of my modded Titan-style GPU blocks — A length of soft tubing that softens and starts kinking ever so slightly as temps rise. It's not kinking against itself, but something I routed it adjacent to. Right now, flow sits at 0.68 GPM (peaks at 0.74-0.75 GPM when not kinked), and I'm pretty satisfied with the temps. I'll wait a few weeks and see where the flow stabilises, and if that value is acceptable, I'll leave it be.

[hihihi8]

dayum... the flow is 0.59GPM now... this is ridiculous.... 2 GPUs + 1 CPU + 1 RAD Block overpowering my D5??

[Limecat86]

20 hours ago, hihihi8 said:

dayum... the flow is 0.59GPM now... this is ridiculous.... 2 GPUs + 1 CPU + 1 RAD Block overpowering my D5??

Have you considered to watercool your GPU's in parallel mode? It will give much less restriction and would still sufficiently cool your cards.  

[hihihi8]

On 11/22/2016 at 9:18 AM, Limecat86 said:

Have you considered to watercool your GPU's in parallel mode? It will give much less restriction and would still sufficiently cool your cards.  

I definitely have considered parallel  However, currently the two cards are piped in different ways (the tubing), so I'm like 90% sure the bottom card is like twice as restrictive as the top card, since it's the one that's causing flow issues. Additionally, parallel would mean redoing 1 or 2 of the more annoying bends of the build, as I can't use standard SLI fittings as you can tell from the pictures. I would have to use a 3-way on the bottom card inlet to pass through to the top, and then flip it over for the outlets, not to mention redoing the piping inside the cards so that they use the same ports as inlet (right now one card uses the outer port as inlet, while the other card uses the outer port as the outlet). Anyways thanks for the thought, I'll probably redo the bottom card's piping on the scheduled next maintenance and give it a try!

 

EDIT: oh woops, this post doesn't have the pictures, anyways, they're modded cards that don't use a full cover block so the bridging of the waterblocks is pretty limited.

[DutchTexan]

1 hour ago, hihihi8 said:

 

I don't know PC water cooling, but I know process loops. Everything I have read indicates there is some form of build up. Excluding the kink in the line you just talked about, it has to be build up in the line, pump impeller damage, or bacteria in the water, building up somewhere in the loop. Higher pressure equals less flow. Lower pressure increases flow. Does your loop have some kind of filtration that you can clean? Do PC loops usually have a filter? There must be some impurities building up in your lines or water blocks AND/OR those impurities are eating away your pump impeller. Assuming you know there are no physical restrictions, kinks.

 

Air bubbles would register as a process temperature or pressure delta. Troubleshooting that is relatively easy. Since you can see the bubbles in your loop, then it is even easier, so I won't explain how to find air pockets. I don't think the air pocket will be a problem, unless you are concerned about temperatures OR that air is making it to your pump. If the air is anywhere near your pump, then I would be very worried. Last thing, the atmosphere may not be sterile, so potentially you could have some kind of bacterial growth in the loop, eventually. Are you using anything other than distilled water? I know food coloring includes sugars and other chemicals that can react and collect in certain areas You probably already know everything I have said, though  

 

If the temperatures are not a concern, then an air bubble shouldn't be that big of a deal. Pressure wise, a pocket of air will affect the loop, but not in a specific rate compounding over time, like what you are explaining. The air bubble should be relatively consistent, unless you have a leak or spill, which you would find out real fast in your PC. Higher temperatures could affect the bubble, expanding the gas, changing the differential pressure across your radiator. Although, that would not explain losing pressure at a constant rate, day after day. Unless you have some kind of kink or a bad seal/connection, everything should be good. So, if temps are gravy and you know air can't get to the pump, don't worry about the air pocket.

 

There's gotta be some shit in your water :1

 

 

 

[hihihi8]

1 hour ago, DutchTexan said:

Excluding the kink in the line you just talked about, it has to be build up in the line, pump impeller damage, or bacteria in the water, building up somewhere in the loop.

Hey! thanks for the reply! I am using pure De-Ionized water. There is a filter in the loop, and I know my radiator is dirty even after days of flushing, so debris build up is definitely contributing to the flow decrease, however I believe the flow is decreasing at too astonishing a rate to be purely from clogged filters, as the flow decreases even after starting up from a overnight shut down (eg. 0.73GPM at shutdown at night, to find it at 0.68 in the morning). As for bacteria, I am using a strip of pure silver in my loop (I think I don't have enough silver, but should still be enough to prevent much growth).

 

1 hour ago, DutchTexan said:

I don't think the air pocket will be a problem, unless you are concerned about temperatures OR that air is making it to your pump.

Temperatures are very good even at these low flow rates i'm having right now, but the decreasing flow is really getting on my nerve lol  Some tiny air bubbles occasionally make it in the pump, doesn't really make a difference or damage anything, as the big bubble is quite far and well-stuck to go anywhere.

 

1 hour ago, DutchTexan said:

a pocket of air will affect the loop, but not in a specific rate compounding over time

Exactly what was troubling me for days lol. It seems to have stabilized (FINALLY!) at 0.58-0.59

 

1 hour ago, DutchTexan said:

Higher temperatures could affect the bubble, expanding the gas, changing the differential pressure across your radiator.

Very interesting point! Explains why my flow seems to change a little bit with temps, During the first 30 seconds after startup, the flow is as low as 0.55GPM, but as I let it idle or do some work, It stabilizes to ~0.58-9

 

1 hour ago, DutchTexan said:

There's gotta be some shit in your water :1

Comes out tasting like copper (Licked some during a drain)