90* Elbows hurt watercooling?

[elrodvoss]

I know that you are never suppost to bend a hose to a point where it kinks, but digging around on various watercooling supply websites and I have seen some of the 90* elbows that you can place to get your line into those hard to reach places.

For me, the biggest thing about my case is I *hate* the rats nest of wires inside. To a point where I would want to mod out the wires to the exact length or find places to hide them (I LOVE the Voodoo wire jobs). I would like to be able to do my water cooling in the same fashion. Have straight lines of tubing and not have tubing curved left and right and just looking "cluttered".

My only consern and question is:
Does using the elbows hinder the flow of water by a recordable amount? An example would be if I put in an elbow, would that reduce my GPH (gallons per hour) by 1-2..or by like 10+?

The pump im looking at is the Swifttech MCP355 and its rated at 120GPH. So losing 1-2 aint bad, but if I lose 1-2 for ever elbow in there, then I might have some issues.

Thanks for any info you can give me on this.

 

[jitpublisher]

You know, that is a good question and I am sure many are wondering the same thing. A perfect chance to contribute some very useful info back to the forums!

Fill up your bathtub with water to a marked level.
Hook up your pump and run a line from the tub to the sink.
Be sure to put your radiator and cooling blocks in the line as well so you will get a fair likeness of the system being used as intended.
Run the pump and time how long it takes to empty the tub.
Now, put a half a dozen elbows in the line, do the same thing again.

Let us know the outcome.

 

[duthoy]

ok, about the 90° elbows;
i did the drawing and planning for a company who builds big warehouses.
when i did the water evacuation, i never used 90° turns, only 2x45°
90° has more impact on the flow than 2 x 45°
i know, we used tubes with like 2 ft. diameter but, basicly it's the same.

 

[whespe]

the simple answer is YES, 90 degree elbows will restrict flow, but the amount of which is negligible for your application.

The loss is cause by turblance when then water turns the corner, known as head loss.

with a bit of reading you can calculate the exact amount, but im going to go ahead and tell you it will only be a few mL per elbow, after all, conservation of mass tells us M.in=M.out

http://en.wikipedia.org/wiki/Head_(hydraulic)
http://www.reefcentral.com/calc/hlc2.php

 

[Human1]

This guy's right. A 90deg elbow does add some friction to the system, but not enough to make a big difference here. According to one of my textbooks, a 90deg turn is about 10 times less of a contributor to friction than a globe valve in wide open position.
Take home message: Put the 90deg turns in, won't hurt anything.

 

[piratepast40]

The main reason that you don't want to kink a hose is that it not only introduces the flow losses from the curve but the kink also reduces the pipe diameter.

A rough rule of thumb for 1/2" diameter, smooth wall pipe is that a 90 degree bend will result in about 1.5 times the flow loss of 1 foot of pipe. A 45 degree bend will result in roughly half that. That's at 1 gpm. You can quadruple that for the 4 gpm pump you have but I suppect that total flow after all losses (heat exchanger, piping, fittings, etc) is more like 2 gpm. If that's the case, just double the loss values instead of multiplying by 4.

As whespe said, it's negligable. Friction, turbulance, and acceleration losses do add up but in such a small system with the 4 gpm energy source you described, the noise from turbulence of 90's vs 45's might be more significant.

It would be interesting to see a performance curve for that pump to see how it reacts to static pressure losses. Again, as whespe said, conservation of mass and energy tells us that energy in has to equal energy out minus losses.

 

[tool_462]

This guy's right. A 90deg elbow does add some friction to the system, but not enough to make a big difference here. According to one of my textbooks, a 90deg turn is about 10 times less of a contributor to friction than a globe valve in wide open position.
Take home message: Put the 90deg turns in, won't hurt anything.

True, and Madxxxracer @ Xtremesystems wrote in a post (he is very knowledgeable in water cooling) that he tested the average flow rate loss of using a simple flow meter compared to 6 90 degree elbows and the loss was roughly the same. People are not afraid of sticking a flow meter in the loop but are worried about elbows, apparently this shouldnt be the case.

 

[elrodvoss]

Thank you everyone for you input.

I remember hearing somewhere that elbows are bad and I was always curious why people would go to such extreams with tubes bending and curving, when they could just use elbows to make it smoother.

The only reason I can thing of, is with each elbow, that just adds another spot that you might have a leak/failure.

 

[Flying-Q]

A thought from the fishkeeping world.

I use plastic tubing for for my filtration system and every tight bend is created by placing the tube into an external preformed shape which is a 90* curve rather than a 90* bend. The tube is not crushed so flow is not restricted. If your look on fishkeeping supplies sites you will see them from manufacturers such as Ehiem.

Good luck

Q

 

[jitpublisher]

Here is a perspective that no one has mentioned yet.

Water flowing too FAST through a system can cause a loss in cooling potential as well. Back in the day of shade tree car mechanics, it was a pretty common deal to replace your thermostat in you car's cooling system as part of a regular tune-up, along with plugs, points, condenser etc. A thermostat is a diaphram type valve that is heat sensitive. It sits in the upper water hose between the radiator and engine block, that stayed closed until your engine reached a certain temperature then opened to allow water to flow- effectively regulating the temperature the engine runs at.

Sometimes this thing would go bad and stick, not opening up allowing water to flow and the engine would overheat. Instead of replacing, sometimes a person would simply take it out completely and run without one. On most cars this simply meant the engine would run very cool, never warm up to proper operating temperature, and your heater inside the car would only blow cold air. BUT on some vehicles, the water would circulate too fast for the radiator to do an effective job of cooling the water, thus, the engine would still overheat- even though the water was flowing much faster though the system without the thermostat in place. An old farmers cheap trick was to stick a big flat washer inside the housing where the thermostat was. The smaller hole in the middle of the washer would slow the water flow down, allowing the radiator time to do it's job cooling the incoming hot water.

What has this got to do with PC water cooling? Probably nothing at all.

But a little restriction by using elbows in the small amount of tubing you are going to use, over distances of only a few inches, or a few feet, will not matter much. If you were pushing the water through a hose 1000 feet long, yes then adding elbows might make a noticable difference in how your cooling system worked.

But all in all, they really won't make much difference. The only real negative I can think of is you are adding more connections, so you have more areas to potentially leak.

 

[BushLin]

There is a small loss but I'd rather have a 90 degree elbow than a really long piece of curved tubing which I would guess creates more of a flow problem and looks terrible. I replaced all my stupid long looped back tubes with short lengths and 90 degree elbows and got much better performance.

 

[Rovemelt]

Adding elbows won't noticeably hinder the flow rate...it should be fine.

 

[rwaritsdario]

I would use lots of 45 degree turns instead, more flexibility to cut down on tubing and less head loss.
With 7/16" ID tubing and metal clamps its really unlikely for a tube to set loose, almost impossible.

Great information on this thread by the way, grats to all knowledgeable posters! 8)

 

[HotFoot]

I figured it would take about 5 mins to do some back-of-the envelope calculations to give you an idea of what penalty you'll pay for using 90 degree elbows in your pipes. Here it is:

Pump: Swiftech MCP355
120 GPH
20.2 ft. head pressure
3/8" tubing

First, I'm going to assume that the flow rate is actually 120 GPH. This represents the worst-case scenario because head loss in a 90 degree elbow is highest for the greatest flow rate. Your actual head loss will be less since the flow rate will be somewhat lower than 120 GPH. Onwards...

Flowrate, Q = 120 GPH = ... = 7.7 [cu.in/s]
Pipe area, A = 1/4*pi*D^2 = 0.1104 [in^2]
Nominal flow speed, V = Q/A = 5.8 [ft/s]

Dynamic pressure = 1/2*rho*V^2 = 0.5*1.938[slug/ft^3]*(5.8[ft/s])^2
Pd = 32.6 lb/ft^2.

The loss from one 90 degree bend is approximately 0.5*Pd, so the head los PER bend is going to be about 16.3 [lb/ft^2], or 0.26 ft of head. This means that for each 90 degree bend, you can expect to lose only about 1% of flow.

This is the first-order approximation. Nonlinear factors will mean diminishing penalties as your flow rate is reduced, but then you do want to keep a certain flow rate, so if you use more 90 degree bends, you'll eventually want to get a stronger pump.

Conclusion: You'll be just fine putting in a few 90 degree joints in your pipe. It might actually be beneficial to put one of these bends just upstream of the CPU or GPU waterblock, as the increased turbulence is beneficial to heat transfer.

Cheers

PS: For a little comparison, the Swiftech Storm waterblock is rated for a 6 psi pressure drop at the 120 GPH flow rate. This is equal to 13.8 ft of head, or 53 90 degree joints.

 

[piratepast40]

HotFoot,
Nice to see someone actually using Bernoulis' equation 8) (albet a simplified version of the concept). What about temperature and pressure coefficients - LOL - just kidding! Amazing how so many things come down to simple principles :idea:

Good job on your "back of the envelope" calculations.

 

[Raviolissimo]

there's a guy named Scott (that's his last name) who worked at Varian on high power radar tubes, and took LOTS of notes, and curve fitted his findings, and wrote a book that is a real good mechanical engineering reference.

to make a long story short, a 90 degree bend like that constitutes a major resistance to flow, but since there are a lot of other turns in the (thermal) circuit, the overall effect will be minor, <5% for sure.

 

[piratepast40]

Is this horse dead yet? :lol: :lol: :lol:

 

[elrodvoss]

Is this horse dead yet? :lol: :lol: :lol:

To be honest, im impressed that I got soo many answers from what I though was a "simple" question. Plus plenty of data to back everything up.

 

[rubix_1011]

This is pretty simple to test - I have the gear to test it out. How many elbows do you want tested over how many feet of tubing? Now, we're talking PVC 90 elbows which are typically a little bit larger in diameter than normal WC fittings, but margin of error is still probably less than 10-20%. We're talking tubing+elbow restriction only, here...so we'd rule out blocks, radiators, etc.

If you're wanting something like 20-30 elbows tested, I don't have that many laying around, but in the neighborhood of 10 or so, I have it covered.

 

[advancedmixedgaming]

This is pretty simple to test - I have the gear to test it out. How many elbows do you want tested over how many feet of tubing? Now, we're talking PVC 90 elbows which are typically a little bit larger in diameter than normal WC fittings, but margin of error is still probably less than 10-20%. We're talking tubing+elbow restriction only, here...so we'd rule out blocks, radiators, etc.

If you're wanting something like 20-30 elbows tested, I don't have that many laying around, but in the neighborhood of 10 or so, I have it covered.

Almost 9 years late bro, but do it for science!

 

[rubix_1011]

Yeah, I realized that we had some spam links in there which brought the thread to the top...and I didn't look at the previous last dates after I removed the spam. My fault guys...closing this one out.