The impact of tubing sizes

[spookmineer]

Ok, if we're being pedantic, there's pressure that is less than atmospheric pressure before the pump inlet, causing the atmosphere to push in and attempt to crush the tubing.

There have been numerous posts from various people where they've shown that with soft-walled tubing, all you need to do is to slightly restrict the run into the pump inlet and the tubing will collapse. Sticking a restrictive block before the pump inlet will achieve the same effect.

If you don't believe me, I'll set up a test and video it in effect for you, using a DDC-2. Believe me, it happens.

It wasn't meant to be pedantic...

Thanks for the explanation. I don't have a watercooling setup (yet) but your post is very plausible, no need for a video.

If anything, my post was to question at how much head pressure there could be a problem, as quick fittings don't (potentially) cause problems just before the pump, but even before this:

I'm only wondering at what amount of head pressure it's not wise to use push fittings anymore.
Is 2 ft too low, is 5 ft good enough?

[edit]
I hope I'm correct here, 1 bar equals 10.2 meter of water pressure (= 33.46 feet).
I guess manufacturers give out specs of overpressure? Assuming that's the case, then people who experience collapsing tubes are having more pressure loss through their loop then the specs given by manufacturers... (going below 1 bar)
Their loop is more restrictive then the pressure from specs can handle, and the pump intake is indeed below 1 bar.

Or Gabe's explanation is the cause of the problems with push fittings:

Negatives in a nutshell:
Safety
- push-in fittings were designed for applications where pressure is <2-7> bars. They were never intended for low pressure environments.

[/edit]

[gabe]

With the amount of headpressure pumps have, I don't think there's actually a vacuum at the suction side of a pump.

There is a depression at the pump inlet, this is a fact, an observation if you will, not an opinion, and not debatable.

This depression causes vinyl tubing to collapse over time upstream of the pump inlet. This is also a fact, an observation, not an opinion thus not debatable either.

A solution to prevent tube from collapsing is smartcoils, but this does not address the sealing issue in quick connects.

Quick connects are only designed to work under positive pressure as specified earlier. The depression causes seal failure. Air will engulf at the seal. This is also an observation, not an opinion. Please take my word literally: I have OBSERVED this phenomenon in many systems from customers, employees, and my very own.

Please learn from my mistakes.

[spookmineer]

There is a depression at the pump inlet, this is a fact, an observation if you will, not an opinion, and not debatable.

I stand corrected (my assumption "The actual lowest pressure (at the suction side of a pump) might still be 2 ft" was false).
But I did not say there wasn't a depression at the pump inlet, I said it isn't a vacuum... ("going below 1 bar" is not the same as a vacuum).

In other words, "Their loop is more restrictive then the pressure from specs can handle, and the pump intake is indeed below 1 bar." is not an exception but it goes for all systems?
Or does depression simply mean, the location in the loop with the least amount of head pressure, but still more then 1 bar?

This depression causes vinyl tubing to collapse over time upstream of the pump inlet. This is also a fact, an observation, not an opinion thus not debatable either.

...? In a way, you are repeating my quote of you

Quick connects are only designed to work under positive pressure as specified earlier. The depression causes seal failure. Air will engulf at the seal. This is also an observation, not an opinion. Please take my word literally: I have OBSERVED this phenomenon in many systems from customers, employees, and my very own.

Please learn from my mistakes.

I believe you! If push-in fittings were designed for applications where pressure is <2-7> bars, it has no place in hardly any watercooling system (it needs 66 feet of head pressure??), with most pumps the push fittings would work below system specs.

Somehow I get the feeling my post is misread, I can't see where I am arguing/denying with anything you posted.

After reading a lot of threads (here and elsewhere) I still know almost nothing about this subject. I'm just curious about it, and perhaps annoyingly so.

[gabe]

I'm just curious about it, and perhaps annoyingly so.

it is a very good thing to be curious :-) yes I am mildly ennoyed -in fact concerned would be more accurate a description, concerned that ppl are not getting the message accross.. but don't worry it's curable with a good glass of wine!

[Turtle 1]

First off Cathra your quit a guy . I am glad to see you finely straightened this miss out. I have been A small tube guy myself for ever when it comes to pc's . Many of us took endless flaming for our stance that small tubing was = to large tubing. As you illustrated so well each has its place and it does infact have everthing to do with components used in the entire loop.

Since so many listen to you I wish you would go over once more the differances between running 2 pumps in series vs. 2 pumps in parallel .

Parallel when talking about pumps is so much superior to series thats its almost laughable anyone who would argue the point of serial being superior in the world of pc usage. Needs to be straightened out also seems to me Cathra is the man for that job.

[Turtle 1]

Perhaps you might want to re-read the first post and take a gander at the pretty pics. Actually he sees the sweet spot at 5/16" (8mm) ID because the 5/16" with a push-fit has essentially the same graph as a 3/8" with a hose barb. And in actuality the difference between the worst and the best is only a measly 1.11 C. I don't think a 1.11 C difference is going to change much in the real world.

TN I agree with what Cartha has shown here but your taking what he said and has shown. Way out of context . Everthing Carths has shown us is loop dependent. With all the 90's your using you have infact a very restrictive system that doesn't meet the spec's of Cartha and the temp variations that he showed would in know way apply to your system as shown. Now if you used copper tubing with straight fitting and nice 90degree bends I would agree with you . I have linked sources were you can buy these fitting . I suspect tho . Now that cats out of bag you will be able to buy in PC shops shortly.

[Cathar]

Parallel when talking about pumps is so much superior to series thats its almost laughable anyone who would argue the point of serial being superior in the world of pc usage. Needs to be straightened out also seems to me Cathra is the man for that job.

Before I launch into it, and I have covered it in the past, may I first enquire as to which two pumps you had put together in parallel, and against what sort of system restriction, that leads you to make that statement with such conviction?

[shiznit93]

Is it possible to use 7/16'' barbs and stretch the 3/8' tubing over them? That should yield flow close to that of using quick-fits right?

[Turtle 1]

Ok I can do that for ya.

I use 2 differant sets ups for this. One set up uses RD30 single pump .
2x. Alphacool AP1510 outlet to y block 2 1/2 inletts and 1 3/4 out let . this go to 4 160's that have been modded to 3/4 id tubing . From the last 160 tubing goes to this part on the inlet side.
http://www.emiplastics.com/water_regulators2.htm

I use both the WO98 or the WO99 . The W099 is intended to be used with skulltrail when released.

From the flow control out let we run a 3/4 tube back to another y Block than the 2 1/2 " lines go to inlets of the 2 pumps to end the 6 loop set up.

Of course with the RD30 no y blocks are used. Both pump set up flow around 8gpm.

[Cathar]

Okay, so if I understood that right, the only real source of restriction you have in the system are the PA160's radiators?

In that instance, yes, pumps in parallel would be significantly better in very low restriction setups like the one you just described.

For a full system though, with a waterblock offering some measure of real restriction, things rapidly start to favor pumps in serial.

For example: Apogee GTX, PA120.2, 2 x Laing DDC with PTS tops

My apologies to the colour blind:



In all cases, pumps in series win. The reason is because of the waterblock. Notice how the series and parallel pump lines intersect at 11LPM (~3gpm). i.e. if your flow rates are above 3gpm, then having the pumps in parallel would be better (for that particular pair of pumps).

[Turtle 1]

My mitake I use the RD 20 . Been toying with use of rd 3o because it has the flow i want 10gpm but it also has some thing i really don't want. Added noise but the use of 2 allows me 10GPM witch I want . Dave just told me I made mistakes above as 2 RD20 are required and y blocks are in fact used sorry about that . But we preferr the Alphacool pumps as their more versatile for our use thats the plus . The minus is they require more room thats the - . Insid the lian li caseits not a problem as far as space is concerned, As we completely redid the cas interior to fit our needs . As well as the face plate and side panels.

[Turtle 1]

Okay, so if I understood that right, the only real source of restriction you have in the system are the PA160's radiators?

In that instance, yes, pumps in parallel would be significantly better in very low restriction setups like the one you just described.

For a full system though, with a waterblock offering some measure of real restriction, things rapidly start to favor pumps in serial.

For example: Apogee GTX, PA120.2, 2 x Laing DDC with PTS tops

My apologies to the colour blind:



In all cases, pumps in series win. The reason is because of the waterblock. Notice how the series and parallel pump lines intersect at 11LPM (~3gpm). i.e. if your flow rates are above 3gpm, then having the pumps in parallel would be better (for that particular pair of pumps).

Thanks so very much for your time. Dave is here helping me now so as not to screw up again . Dave said there is only one part that he has ever used that would cause alarms to go off if someone put a system together using the wrong parts. He says you are responsible for that part . The storm he says congrats on a great part design.
But thats part isn't used in the high flow system . We have our own part for that now.

Dave said that inorder to get 8 gpm flow and use parts that turn off and on with the PC he has to run 2 pumps in parallel to achieve this he is not happy with the noise but we have gotten rid of most of it to the point its just a slit hum . but 2 RD 20 are noiser and 2 rd30 are worse .
Hopefully someone will come out with a better hi flow pump.

[Cathar]

Some questioned the correctness of old beliefs. Here's an analysis on an old-school system.

The system we're comparing is a DangerDen Maze 4, an Eheim 1048 or Eheim 1250, and a Thermochill HE120.2 with some decent fans on board. ~4 years ago, this was a high performance system in the USA.

Interactive flow-performance results, on an 80W mildly overclocked CPU of the same era may be found here.

The HE120.2 radiator performance curve can be found here. We'll be using the pink curve to represent the fans, which would be about the full-time tolerable noise limit for many individuals.

We'll be assuming a 120W CPU heat load though, which was more typical of the high-end overclockers of the day. Pump heat dump is ~10W for the 1250, and 5W for the 1048.

The flow rates between 3/8" and 1/2" are seen on this graph:



For the 1048:

3/8" => 4.8LPM
1/2" => 6.9LPM

For the 1250:

3/8" => 6.2LPM
1/2" => 9.6LPM

Looking to our Procooling graph, and multiplying the values by 1.5 (120/80), we get:

1048 1/2" to 3/8" delta => 1.4C
1250 1/2" to 3/8" delta => 1.8C

The estimated radiator water temperatures are:

For the E1048:
125W @ 4.8LPM => 5.4C
125W @ 6.9LPM => 5.2C
=> 0.2C delta

For the E1250:
130W @ 6.2LPM => 5.5C
130W @ 9.6LPM => 5.2C
=> 0.3C delta

With lower speed fans, the radiator performance deltas get slightly larger.

So the total delta for 3/8" tubing with barbs to 1/2" tubing with barbs is:

1048 => 1.9C
1250 => 2.1C

Those were the differences for just moving from 3/8" to 1/2", for the really heavy overclockers. Obviously that is what justified the move for many.

Like I've said for those who were wondering why things are different nowadays. Things have changed. Water-cooling has evolved.

[ranker]

Some questioned the correctness of old beliefs. Here's an analysis on an old-school system.

The system we're comparing is a DangerDen Maze 4, an Eheim 1048 or Eheim 1250, and a Thermochill HE120.2 with some decent fans on board. ~4 years ago, this was a high performance system in the USA.

Interactive flow-performance results, on an 80W mildly overclocked CPU of the same era may be found here.

The HE120.2 radiator performance curve can be found here. We'll be using the pink curve to represent the fans, which would be about the full-time tolerable noise limit for many individuals.

We'll be assuming a 120W CPU heat load though, which was more typical of the high-end overclockers of the day. Pump heat dump is ~10W for the 1250, and 5W for the 1048.

The flow rates between 3/8" and 1/2" are seen on this graph:



For the 1048:

3/8" => 4.8LPM
1/2" => 6.9LPM

For the 1250:

3/8" => 6.2LPM
1/2" => 9.6LPM

Looking to our Procooling graph, and multiplying the values by 1.5 (120/80), we get:

1048 1/2" to 3/8" delta => 1.4C
1250 1/2" to 3/8" delta => 1.8C

The estimated radiator water temperatures are:

For the E1048:
125W @ 4.8LPM => 5.4C
125W @ 6.9LPM => 5.2C
=> 0.2C delta

For the E1250:
130W @ 6.2LPM => 5.5C
130W @ 9.6LPM => 5.2C
=> 0.3C delta

With lower speed fans, the radiator performance deltas get slightly larger.

So the total delta for 3/8" tubing with barbs to 1/2" tubing with barbs is:

1048 => 1.9C
1250 => 2.1C

Those were the differences for just moving from 3/8" to 1/2", for the really heavy overclockers. Obviously that is what justified the move for many.

Like I've said for those who were wondering why things are different nowadays. Things have changed. Water-cooling has evolved.

I think most of us here at XS grasp the concept and understood how things have evolved and changed.

However, I feel sorry for the fight you're getting over on [H]. Seems like they're a little slow and trying to deliberately spin your results into championing how low-flow setups have been superior all along. (For those that are interested in seeing what happens when you anger the natives: http://www.hardforum.com/showthread....1200495&page=3)

I'm sure most people here at XS will appreciate this graph as it helps demonstrate your conclusions by allowing us to compare how it applied back then and how it applies to today's water cooling components.

[LogAn'sRun]

I don't like the natives, I'm happy right here. . . .

[septim]

[MotF Bane]

However, I feel sorry for the fight you're getting over on [H]. Seems like they're a little slow and trying to deliberately spin your results into championing how low-flow setups have been superior all along. (For those that are interested in seeing what happens when you anger the natives: http://www.hardforum.com/showthread....1200495&page=3)

Would that be considering angering the natives, or being intelligent in the proximity of a troll home?

[sick_g4m3r]

Excellent findings!

[ranker]

Would that be considering angering the natives, or being intelligent in the proximity of a troll home?

Being intelligent in a proximity of a troll home for sure.

[RickCain]

Being intelligent in a proximity of a troll home for sure.

You linking to another forum and your comments could be considered trolling too......

But hey, anything to keep the flaming right?

[migueld]

You linking to another forum and your comments could be considered trolling too......

But hey, anything to keep the flaming right?

Heh so true, that's pure flamming/trolling. It's sad

[ranker]

You linking to another forum and your comments could be considered trolling too......

But hey, anything to keep the flaming right?

Heh so true, that's pure flamming/trolling. It's sad

Hi Pot, Meet Kettle, You're black too.

If anything I was thanking Cathar for posting his results so it could be more easily understood/referenced during discussions and pointed out that he wasn't so lucky to receive the same reception at another forum.

[SiGfever]

Considering that Cathar has probably forgotten more about water cooling than most of us will ever know who cares what the uninformed think?

[Top Nurse]

Being intelligent in a proximity of a troll home for sure.

Yeah, I get kind of itchy whenever I come over to XS as there are lots of trolls here as well. It's okay because they do the same dog and pony show, right?

[TRG]

Great read and great information!

I personally prefer the small 5/16" tubes in combination with quick-fit fittings because they're small, have a great name in industry use (pneumatics / professional watercooling) and are very easy to use.