• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Lets Settle this once and for all: Water Cooling and Hose Size

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

GreenJelly

Member
Joined
Oct 16, 2006
I really want to learn, and come up with the truth. I have done expiriments on my case and certain Hose's.

Now, there is NO dispute that 1/2 ID hose is less restrictive then any smaller ID hose.

The problem I have is that I beleive, and I may be wrong, that the smallest hose size or restriction will ultimately be the weekest link in the chain. This makes sense only because unlike a gas, water doesnt compress.

Therefor in theory, based on the above theory; you could run the following expiriment.

Requirements:
Pump Size, Voltage, ube length and Tube loop height must remain constant. The Pump and Hose must be completely full, and pump running before any measurements are takin. The Pump must be extremely powerfull, with high flow and high preasure.

Experiment Procedure
We use 3 different hoses. Hose "A" is a 1/2 ID, Hose "B" is a 1/4 ID, and Hose "C" is a 1/2 ID with a reduced section measuring 1/4 ID.

Each expirement would require a Fill Bottle and a Drain Bottle. Each Hose (A-C) would undergo the same expiriment. The pump would be turned on, and would fill the fill bottle until all bubbles are removed. It would be marked and a Time would be taken. After 60 seconds, the waterlevel marked again.

Measure the difference and you get the flow rate of each hose over one minute.

Expected Result: Hose A would have the highest flow rate. Hose B would have the secondhighest, but by very small difference bettween Hose C

Lastly The expirement should be repeated with a lowflow high pressure pump.
In this case I would expect all 3 Tubes to remain constain in flow rate.

What do you guys think, and please provide resources, or if you can, complete the expiriment. I got to run, I will provide details of my experiments results, with you later today.

Edit: Definitions for the Noobies
OD = Outside Diameter, size of the hose on the outside.
ID = Inside Diameter, size of the hose on the inside.
Compression Fittings = Hose Clamps that have small barbs, and screws that tighten around the hose to compress the tube from falling off. Always steal or metal
Barbs = Plastic or metal parts that slide into the tube, and that have a ridge that prevents the hose from sliding off. Don’t be fooled, always use clamps to secure your barbs. Non Secured Barbs will probably leak after some time.
Clamps = "C" metal clamps found at home depot, which are made out of steal and that are tighten with a screw driver. Over tightening C clamps can break plastic barbs, or can damage hose over time. Plastic Quick clamps work well and are also a valid product. Prefered for some hose sizes, they are easily applied, are non-conductive and are safe and can help prevent from over tightening that occurs with C clamps.
 
Last edited:
The question of "best" hose size will never be settled, too many folks have fun arguing about it :) And, there are good arguments for both high flow and low flow systems.

My personal opinion is that for optimum performance the system should match. If you use blocks with 1/2" barbs you should use tubing with either 1/2" or 7/16" ID (the inside diameter of most 1/2" barbs is actually ~3/8" so the 7/16" ID tubing isn't restrictive). If you want to use blocks with other sized barbs, use the same criteria in choosing the tubing.

And, it really doesn't make sense to use the same hardware and just swap out the tubing. If you are building a high flow/high pressure system you'd choose parts appropriate to that system. If you're building a low pressure/low flow system you'd choose different parts. A water cooling system is a SYSTEM. Arbitrarily mixing and matching parts designed for different purposes could easily get you the worst of both worlds.
 
Quote:

"This is from Cathar

Quote:
Been rolling the whole tubing size idea around in my head, and thinking about trade-offs and the like.

I like the idea of 3/8" ID tubing, but I just can't shake the feeling that for >4LPM that it starts to become an increasingly significant source of restriction for those who wish to make use of strong pumps capable of pushing the higher flow rates. It's not that 3/8" tubing is bad at all for coping with moderate flow rates, it's just that it could be better. For example at 6LPM, 7' of 3/8" ID tubing is offering pretty close to 1mH2O of pressure drop all by itself.

But 3/8" is attractive because it's very light, and it bleeds air-bubbles fast.

1/2" tubing is fat and unattractive. Unless flow rates are getting past the 6LPM mark, air-bubbles don't bleed very well. It's heavy, and it requires fairly thick walls (1/8") before it can turn good radii without kinking, but this wall thickness comes at a cost of making it stiffer to turn, thus putting more leverage on the water-block's all important thermal contact. However, it takes around 13.5LPM before 7' of 1/2" ID tubing offers 1mH2O of pressure drop, so really it's almost overkill.

So I looked to the middle-ground, that being 7/16" (~11.1mm) that has 3/32" wall thickness for a total of 5/8" OD. Per length of tubing it's about 2/3's the weight of the 1/2" ID (3/4" OD) tubing. Being a thinner ID it is able to be bent into tighter radii without kinking, allowing for the use of the 3/32" wall thickness, which means that it also becomes easier to turn those radii. It offers 1mH2O of pressure drop at 9.5LPM for a 7' length, which pretty much puts it still as a very attractive offering.

Then I took into account stretching the 7/16" ID tubing over 1/2" OD fittings (barbs) with 10mm ID orifices. Due to the "lip effect" the 1/2" ID tubing actually offers nearly 3x the transitioning resistance at fittings as the 7/16" tubing whose ID more closely matches the ID of the fitting. Over a typical full system when fitting resistance is taken into account, the 7/16" ID tubing offers almost the same amount of tubing + fitting resistance as the 1/2" ID tubing.

Results were obtained using the pressure drop calculator from http://www.pressure-drop.org.

This all got me to thinking that really what us 1/2 inchers may really want to be doing is fitting 3/32" thickness walled 7/16" ID tubing over our 1/2" barbed systems, and pretty much be enjoying no extra system resistance, but gaining the benefits of lighter tubing that is easier to bleed (bleeds very well at a predicted ~5LPM), easier to bend, isn't as bulky, "hangs" less off water-blocks, and is significantly cheaper due to less wall material being used."
 
So what's the debate? Ya know which one is going to net the largest flowrate, even though the open system tests might not follow the exact same curves as a closed system test would.
Most of your restriction is still going to be at the barbs and inside the waterblocks, because that's your smallest cross-sectional areas with the sharpest transition.

Transitions are going to be the greatest achiles heel. If you use a regular 1/2" - 1/4" barb C will be your loser because the ID of the 1/4" end is actually smaller than the 1/4" tubing it's feeding, and the transitions are blunt.

If you were to smooth out the transition with a 3" long tapered tube, I'd bet your C example would beat B, but that would require a custom fitting that most cannot make, so the data gained would still be only academic.
 
SiGfever said:
Quote:

"This is from Cathar

Quote:
Been rolling the whole tubing size idea around in my head, and thinking about trade-offs and the like.

I like the idea of 3/8" ID tubing.........

This all got me to thinking that really what us 1/2 inchers may really want to be doing is fitting 3/32" thickness walled 7/16" ID tubing over our 1/2" barbed systems, and pretty much be enjoying no extra system resistance, but gaining the benefits of lighter tubing that is easier to bleed (bleeds very well at a predicted ~5LPM), easier to bend, isn't as bulky, "hangs" less off water-blocks, and is significantly cheaper due to less wall material being used."

Do you mind sharing the link to the source of this quote? I would like to be able to read the entire thread, as well as refer to it, as needed.

Thanks.
 
Cathar’s comments regarding turning radius is incorrect. All you need to do is review Tygon’s spec sheet. You’ll see that 7/16” x 3/32” tubing has a bending radius of 1 3/4" while 1/2" x 1/8” tubing has a bending radius of 1 1/2".

What you really want is 7/16” x 1/8” tubing. That has a bending radius of 1 3/8”...3/8” smaller than 3/32” walled tubing. 3/8” x 1/8” tubing has a bending radius of 1 1/8”

http://www.tygon.com/media/documents/S0000000000000001013/tygr3603.pdf

As for head loss in tubing 1/8” thick...
The Friction Head of 4 ft of 1/2" tubing, at 2 GPM is 0.58 ft.
The Friction Head of 4 ft of 7/16" tubing, at 2 GPM is 1.141 ft.
The Friction Head of 4 ft of 3/8” tubing, at 2 GPM is 2.5 ft.
The Friction Head of 4 ft of 1/4” tubing, at 2 GPM is 20.58 ft.

3/8" tubing can rob nearly 24" of head from the pump over 1/2" tubing. This is why it is so important to use 1/2" tubing. However, 7/16” is a good compromise if you need the smaller radius but still have room for thicker tubing.

I’m currently building a water-cooled microATX slim case, so I have to use 3/8” tubing for the smaller radius and smaller diameter.

slimcase4.jpg


slimcase5.jpg


slimcase6.jpg


slimcase7.jpg
 
I find that many parts that are sized and come with 1/2" barbs have internals that go way bellow 1/2". My Resivor is an example.

In fact, I live near Frozen CPU, and have looked at almost every single item in their stock. I have to say that this seems to be a norm, though not always, the case. Parts come with 1/2 or 3/8 inch barbs, yet are restrictive WAY below thoose hose size.

I find that the Swiftech storm is VERY restrictive, and it comes with 1/2 and 3/8" Barbs. I find that it makes very little difference to the flow if your entire system is 8mm and only the Swiftech remains at 3/8"...

Very Odd behavior, and I really feel that this is a area that is so underestimated.

I try and pump as much water as I can through my setup. I use 1/2" hoses 3/8" and 8mm. I see very good results without no difference if I start raising my 8mm parts up to 3/8"...

Very Weird, and is why I started this thread. The theory for larger tube is solid... who can argue that 1/2" is more restrictive then 3/8"... But in practice I have tried hard to keep my system high flow, and still end up with a minimum diameter of 8mm. Anything I use that higher then 8mm, is simply due to parts available, and sizes provided by equipment.
 
The issue of 1/2" barbs having slightly smaller IDs is really no big deal when you are talking about using a pump like a DDC+ with 20' of head. 8mm is kind of small as 1/2" barbs usually have an inner id of about 7/16". I don't understand however why you are limited to some parts that only have 8mm barbs.
 
Barbs are subject to a manufacturer's whim.
I pulled 3 examples from my desk drawer:
*DD Polypropylene=10mm
*Home Depot White Nylon=9mm
*Lowe's Brass=8mm (and it's the most expensive of the lot too)

They're mostly used for pressurized applications where restriction isn't as much of an issue. They don't make them just for those of us on the bleeding edge of getting every last drop of performance.

When I make blocks, I use thin walled copper tubing and solder on a brass ferrule (aka compression ring) for a barb. That gives me the largest ID I can get and still not have to force a hose onto them.
I'd used the same method on a brass fitting before, by drilling it out through the threaded end. The old barb is replaced with a hole that accepts the new copper tube soldered into place. Though you couldn't do this as easily using 1/2" tube and a 1/4" Male NPT fitting.
 
Last edited:
You can make your own barbs if you have the skills...

I couldnt find flanged corner 17mm barbs with a BSB1/4 inch tread so I got someone to make em for me.

 
GreenJelly said:
I find that many parts that are sized and come with 1/2" barbs have internals that go way bellow 1/2". My Resivor is an example.

In fact, I live near Frozen CPU, and have looked at almost every single item in their stock. I have to say that this seems to be a norm, though not always, the case. Parts come with 1/2 or 3/8 inch barbs, yet are restrictive WAY below thoose hose size.

I find that the Swiftech storm is VERY restrictive, and it comes with 1/2 and 3/8" Barbs. I find that it makes very little difference to the flow if your entire system is 8mm and only the Swiftech remains at 3/8"...

Very Odd behavior, and I really feel that this is a area that is so underestimated.

I try and pump as much water as I can through my setup. I use 1/2" hoses 3/8" and 8mm. I see very good results without no difference if I start raising my 8mm parts up to 3/8"...

Very Weird, and is why I started this thread. The theory for larger tube is solid... who can argue that 1/2" is more restrictive then 3/8"... But in practice I have tried hard to keep my system high flow, and still end up with a minimum diameter of 8mm. Anything I use that higher then 8mm, is simply due to parts available, and sizes provided by equipment.

Mind posting some pictures of your setup becuase if it is whats in your sig I have no idea why you need those different sized tubing diameters.

Also ultra-high flow will not always be the best option for cooling. Since it is a closed loop the water will spend the same amount of time in each component of the system no matter how slow the water moves. However it takes a fair amount of energy to heat up and cool down water which means you will see a higher temperature differential in slower moving loops but lower friction than in higher flow loops.

The reason you use 1/2" ID is becuase you can either experaince your restriction in your blocks or you can experiance restriction in your blocks and tubing. Since restriction means my pump is working harder and there is more friction in my loop, I think I prefer less restriction don't you?

Almost forgot for those barbafobics out there I found some great stuff from DD.
9/16" barbs http://www.frozencpu.com/products/2...n_916_Thread_Barb_-_12_ID.html?tl=g30c101s178


For your G 1/4" threaded hardware. (I have these on my bip3 and the ID is atleast 1/2" i'll check if i pull apart my pelt loop anytime soon.)
http://www.frozencpu.com/products/2...G_14_Thread_Barb_-_12_ID.html? tl=g30c101s178

Or if your really barbafobic Mcmaster sells 5/8" barbs
 
speed bump said:
Mind posting some pictures of your setup becuase if it is whats in your sig I have no idea why you need those different sized tubing diameters.

On the 3rd Im moving into my own apartment, and will get roadrunner there. I will then resetup my website, and I will put together a detailed description with images of my two builds.

The reason I need 1/8" is that the radiator is close to the back. Its a stacker so it has 2 places for PSU's. The radiator takes up the top PSU slot.

I didnt construct the case, some guys in Germany did, and the pump and water block from my HTPC build came with the setup. The radiator comes almost to the back, streight barbs would run into the support structure of the case and the "box" that contains the USB, the Mouse and Keyboard connectors, etc. Therefor the guys in Germany used a "L" bracket with a 8mm compression fitting for it. I dont have much choice to fix this, as for the case is modified (VERY nicely may I say) for this radiator, and for a bunch of other reasons.

The original setup was a low flow, low pressure setup. (1 l/min). Like I said it was made by germans, and all of the European parts are running lowerflow systems; for the resons you mention bellow.

I would have been very happy with the original setup, but the P5B (basic) was reporting high temps (54-64c). I decided to spend just over $200 for the pump, the storm, and the resivor I use now.

I found out that this only dropped the temp a few degrees, though my flow drastically increased. The high temps were caused by some problems with the p5B basic bios and the X6800 and/or 1066 Corsair XMS memory.

When I switched to Gigabyte (after getting sick of Asus because of the P5B), I saw a 20c drop in temps at idle. Even compairing the Dual Prime95 torture test verse the Asus Idle I reach a 10-15c drop.

I now wish I didnt spend all that money... but hey, I get to watercool my HTPC:)

Also ultra-high flow will not always be the best option for cooling. Since it is a closed loop the water will spend the same amount of time in each component of the system no matter how slow the water moves. However it takes a fair amount of energy to heat up and cool down water which means you will see a higher temperature differential in slower moving loops but lower friction than in higher flow loops.
wisdom is nice... Hence I agree

The reason you use 1/2" ID is becuase you can either experaince your restriction in your blocks or you can experiance restriction in your blocks and tubing. Since restriction means my pump is working harder and there is more friction in my loop, I think I prefer less restriction don't you?
I agree with the restriction of tubing. If there is restriction of your block, radiator, resivor, etc. then it seems to be only valuable to go with the tube size that is less restrictive then this most restrictive part.

Mike
 
Graystar said:
As for head loss ...
The Friction Head of 4 ft of 1/2" tubing, at 2 GPM is 0.58 ft.
The Friction Head of 4 ft of 7/16" tubing, at 2 GPM is 1.141 ft.
The Friction Head of 4 ft of 3/8” tubing, at 2 GPM is 2.5 ft.
The Friction Head of 4 ft of 1/4” tubing, at 2 GPM is 20.58 ft.
I'm curious - where did you get those numbers ...?
 
Graystar said:
Cathar’s comments regarding turning radius is incorrect. All you need to do is review Tygon’s spec sheet.

I think Cathar was speaking of general vinyl tubing (but even then it varies by formulation, home depot v.s. Masterklear v.s. Clearflex nets you big differences) but yes, as far as Tygon goes his numbers are incorrect. Then again, thats not surprising as Tygon is simply awesome tubing, if I could justify it I would use it.

As far as what tubing is best, I don't think there is a single "perfect" size of tubing. You need to use what works best for your situation. I've used 3/4" tubing in systems with tons of space (I think it looks cool which was the major reason why) and 5/16" in very small cases. Both work and that's the important thing. 1/2" tubing would have never worked in the small case I was working in but 5/16" did and it didn't matter what the pressure drop was because I used about a foot of tubing (sorry, I don't have pictures).
 
Back