watercooling 90 degree elbow bend

gotensan01

Golden Member
Jul 6, 2004
1,446
0
0
Okay, this may be an obvious question but I am confused. I've read numerous threads in here about water cooling but am a little confused about the use ot 90 degree adapters. I believe I've read replies saying that the use of these adapters will cause head loss and it won't be good. However if I am not mistaken, I think I've read that having 90 deg bends without any adapters is bad (obvious) but if you use an adapter like an elbow bend, it is fine.

Also, do they sell stuff like 45 degree bend elbows?
 

cirthix

Diamond Member
Aug 28, 2004
3,616
1
76
umm, get thicker tubing. i use reinforced stuff, which is 1/2" id but 3/4" od. it never kinks, even if i try to make it kink. its a little on the stiff side, so i had to use a lot of elbows. i got it at my local ace
 

gotensan01

Golden Member
Jul 6, 2004
1,446
0
0
I have Tygon 1/2" ID, 3/4"OD. I have three areas where the bends are tight enough to the point of flattening to an oval but not kinking. I don't know how bad this flattening is but is it better than having elbows?

cirthix, do you know if having elbows degrades cooling performance? Seeing that you used a lot of them, I would imagine you'd say they don't ;)
 

cirthix

Diamond Member
Aug 28, 2004
3,616
1
76
i have a via aqua 1300 pump, so elbows are no problem for me, i probably have arond 250 gallons per hour at a couple feet of head, so elbows dont cause any real perfomance degradation. if the flattening isnt bad, save the effort of putting elbows and leave it as it is, but if you have a weak pump and the bends are going to a waterblock, you may want to use elbows. bent thick tubing can be like a spring and crush/lift up something, even bending pcb while trying to. i'm pming this to you because the thread is probably too old to find again
 

HardWarrior

Diamond Member
Jan 26, 2004
4,400
23
81
While 90 degree don't cause a problem with flow, they do decrease velocity. As long as you don't have one too close to the intake of a high-velocity block you'll be fine. Since you've got good tubing I would suggest staying away from fittings and just plumb the turns.
 

gotensan01

Golden Member
Jul 6, 2004
1,446
0
0
Originally posted by: HardWarrior
While 90 degree don't cause a problem with flow, they do decrease velocity. As long as you don't have one too close to the intake of a high-velocity block you'll be fine. Since you've got good tubing I would suggest staying away from fittings and just plumb the turns.
Sorry, what does plumb the turn mean? It does make sense that you don't want to have elbow by high-velocity blocks, thanks for the info. I think I'm going to just use 45 degree adapters just to reduce the head loss some more.
 

HardWarrior

Diamond Member
Jan 26, 2004
4,400
23
81
Never apologize for asking a question. :) Make the turn without the fitting is what I mean. The tubing will naturally smooth the turn all by itself. A 90 degree fitting is very abrupt and will actually dimmish head, though not loop flow.
 

prometheusxls

Senior member
Apr 27, 2003
830
0
0
For a 90 Degree fitting, there is a pressure drop associated with the bend and there is also a drop associated with the fitting since there is a limited cross sectional area for flow. The idea is that to turn the flow you need to exert force, if you exert force than you have to put up with additional viscous energy loss. With respect to cross sectional effects in a fitting, the effect is relatively slight if you have a well designed fitting. Imagine the difference between walking through a door frame and walking through a door w/ no frame (a hole in the wall). This is about the level of restriction introduced by a fitting.

Let us take the case of a bent tube. Also there will be a loss due to the turning of the flow of water just as in a fitting. It makes sense since the stream of water is still being bent and this is causing energy loss. However this is a minor effect compared to the actual loss of cross sectional area associated with bending.

If the round tube actually bends into an ellipse then it loses cross sectional area, potentially to a great degree. If the Cross-sectional area loss is sufficient the flow can be cut off completely. This is called a kink. Any loss of cross sectional area will create a drop in pressure (head). Since we are talking about turbulent flow, it is obvious that this loss of cross section will have a much greater influence than the viscous forces.

In real world applications depending on the system this loss of cross section may or may not be significant. If you have a powerful pump with a ton of head but low flow there may not be an issue. If however you have a high flow pump w. low head then loss of cross-section will hamper your efforts greatly. Once you prescribe a path in terms of the number of turns and such, the cross sectional area for flow is the only thing you can improve. Bending the pipe almost always will reduce that area more than a good fitting.

 
Sep 17, 2004
29
0
0
Please allow a chemical engineer to weigh in on the subject, we tend to know quite a lot about flow through tubes. The head loss experienced by a 90 degree bend in a pipe is a function of many things, fluid viscosity, fluid density, fluid flow velocity, pipe diameter, and pipe roughness. Since we are talking about narrow smooth plastic tubing and water at or near ambient the head loss is very very minimal. It is completely incorrect that the velocity through the elbow drops. The velocity in the entire length of tube must be the same or you will accumulate fluid in one section of the tube, which clearly does not happen. With enough information I can calculate the exact head loss you are experiencing and how much effect that has on your cooling, but I'll bet that it is much less than you think. And by the way somebody else mentioned turbulent flow, you can't know whether the flow is turbulent or not without calculating the Reyonlds number or using some sort of vortex detector. Besides, you want turbulent flow in this sort of application, turbulent flow conducts heat much better than laminar flow does, and you want to discard heat through the pipe wall.
 

gotensan01

Golden Member
Jul 6, 2004
1,446
0
0
Originally posted by: StormGod
Please allow a chemical engineer to weigh in on the subject, we tend to know quite a lot about flow through tubes. The head loss experienced by a 90 degree bend in a pipe is a function of many things, fluid viscosity, fluid density, fluid flow velocity, pipe diameter, and pipe roughness. Since we are talking about narrow smooth plastic tubing and water at or near ambient the head loss is very very minimal. It is completely incorrect that the velocity through the elbow drops. The velocity in the entire length of tube must be the same or you will accumulate fluid in one section of the tube, which clearly does not happen. With enough information I can calculate the exact head loss you are experiencing and how much effect that has on your cooling, but I'll bet that it is much less than you think. And by the way somebody else mentioned turbulent flow, you can't know whether the flow is turbulent or not without calculating the Reyonlds number or using some sort of vortex detector. Besides, you want turbulent flow in this sort of application, turbulent flow conducts heat much better than laminar flow does, and you want to discard heat through the pipe wall.
:confused:
brain...oozing out of ear...must convert to more simplistic terms...
 
Sep 17, 2004
29
0
0
Originally posted by: gotensan01
Originally posted by: StormGod
Please allow a chemical engineer to weigh in on the subject, we tend to know quite a lot about flow through tubes. The head loss experienced by a 90 degree bend in a pipe is a function of many things, fluid viscosity, fluid density, fluid flow velocity, pipe diameter, and pipe roughness. Since we are talking about narrow smooth plastic tubing and water at or near ambient the head loss is very very minimal. It is completely incorrect that the velocity through the elbow drops. The velocity in the entire length of tube must be the same or you will accumulate fluid in one section of the tube, which clearly does not happen. With enough information I can calculate the exact head loss you are experiencing and how much effect that has on your cooling, but I'll bet that it is much less than you think. And by the way somebody else mentioned turbulent flow, you can't know whether the flow is turbulent or not without calculating the Reyonlds number or using some sort of vortex detector. Besides, you want turbulent flow in this sort of application, turbulent flow conducts heat much better than laminar flow does, and you want to discard heat through the pipe wall.
:confused:
brain...oozing out of ear...must convert to more simplistic terms...


To put it simply, don't worry about head loss from 90° elbows. Their effect in a situation like yours is purely academic, you should not notice any reduction in velocity through your system for having put them in. Hope that helps.
 

gotensan01

Golden Member
Jul 6, 2004
1,446
0
0
Cool, thanks for your expertice. I was about to take a DD Z chipset block out because the barbs are at 90 degree angles to the block. But I guess I'll keep it in now.
 

HardWarrior

Diamond Member
Jan 26, 2004
4,400
23
81
Originally posted by: StormGod
Originally posted by: gotensan01
Originally posted by: StormGod
Please allow a chemical engineer to weigh in on the subject, we tend to know quite a lot about flow through tubes. The head loss experienced by a 90 degree bend in a pipe is a function of many things, fluid viscosity, fluid density, fluid flow velocity, pipe diameter, and pipe roughness. Since we are talking about narrow smooth plastic tubing and water at or near ambient the head loss is very very minimal. It is completely incorrect that the velocity through the elbow drops. The velocity in the entire length of tube must be the same or you will accumulate fluid in one section of the tube, which clearly does not happen. With enough information I can calculate the exact head loss you are experiencing and how much effect that has on your cooling, but I'll bet that it is much less than you think. And by the way somebody else mentioned turbulent flow, you can't know whether the flow is turbulent or not without calculating the Reyonlds number or using some sort of vortex detector. Besides, you want turbulent flow in this sort of application, turbulent flow conducts heat much better than laminar flow does, and you want to discard heat through the pipe wall.
:confused:
brain...oozing out of ear...must convert to more simplistic terms...


To put it simply, don't worry about head loss from 90° elbows. Their effect in a situation like yours is purely academic, you should not notice any reduction in velocity through your system for having put them in. Hope that helps.

SHOULD not? After all this you can't be definite? I never said that 90 degree HARD turns were inherently bad, only that it isn't a good idea to place one just before a high-impingment block. This you didn't even bother to address.

 

gotensan01

Golden Member
Jul 6, 2004
1,446
0
0
Okay, I think my original question was a bit too general. One of the main reasons why I am asking this question is because I have the DD Z chipset block. Now I'm just wondering if I should get a much better passive heatsink and forget about water cooling my nb. Basically I just wanted to know if the extra tubing (that is somewhat flattened) and the barbs that are at 90 deg angles with the block hinder the cooling to my cpu and gpu.

Simply put, does cooling the nb using the Z chipset block cause enough flow/cooling loss to the rest of the setup that it warrants me to replace the water block with a heatsink?
 
Sep 17, 2004
29
0
0
Originally posted by: HardWarrior
Originally posted by: StormGod
Originally posted by: gotensan01
Originally posted by: StormGod
Please allow a chemical engineer to weigh in on the subject, we tend to know quite a lot about flow through tubes. The head loss experienced by a 90 degree bend in a pipe is a function of many things, fluid viscosity, fluid density, fluid flow velocity, pipe diameter, and pipe roughness. Since we are talking about narrow smooth plastic tubing and water at or near ambient the head loss is very very minimal. It is completely incorrect that the velocity through the elbow drops. The velocity in the entire length of tube must be the same or you will accumulate fluid in one section of the tube, which clearly does not happen. With enough information I can calculate the exact head loss you are experiencing and how much effect that has on your cooling, but I'll bet that it is much less than you think. And by the way somebody else mentioned turbulent flow, you can't know whether the flow is turbulent or not without calculating the Reyonlds number or using some sort of vortex detector. Besides, you want turbulent flow in this sort of application, turbulent flow conducts heat much better than laminar flow does, and you want to discard heat through the pipe wall.
:confused:
brain...oozing out of ear...must convert to more simplistic terms...


To put it simply, don't worry about head loss from 90° elbows. Their effect in a situation like yours is purely academic, you should not notice any reduction in velocity through your system for having put them in. Hope that helps.

SHOULD not? After all this you can't be definite? I never said that 90 degree HARD turns were inherently bad, only that it isn't a good idea to place one just before a high-impingment block. This you didn't even bother to address.


Don't be ridiculous HardWarrior, in the complete abscence of any numeric data and without solving any of the governing equations I can not be definite. The scientific community as a whole generally frowns on speaking in absolutes, especially when one is not absolutely sure. I would be very surprised if after going through all the calculations the resulting head loss was anything but trivial but I can not be certain until I go through the excercise. To say otherwise would simply be lying. Does anyone here really want to go through the trouble?
 

HardWarrior

Diamond Member
Jan 26, 2004
4,400
23
81
There's nothing ridiculous about expecting something more than a guess from a person who introduced themselves as THE definitive source of fluid dynamics truth. I may be wrong here, but I don't think anyone said that head-loss was an issue with hard 90 degree turns. Only that velocity may be an issue if the turn is placed just prior to a block.

I'm not trying to put you down, but even without hard numbers I expected more, rightly or wrongly.

"Please allow a chemical engineer to weigh in on the subject, we tend to know quite a lot about flow through tubes."

This is fairly heady stuff, SG.
 

gotensan01

Golden Member
Jul 6, 2004
1,446
0
0
Originally posted by: gotensan01
Okay, I think my original question was a bit too general. One of the main reasons why I am asking this question is because I have the DD Z chipset block. Now I'm just wondering if I should get a much better passive heatsink and forget about water cooling my nb. Basically I just wanted to know if the extra tubing (that is somewhat flattened) and the barbs that are at 90 deg angles with the block hinder the cooling to my cpu and gpu.

Simply put, does cooling the nb using the Z chipset block cause enough flow/cooling loss to the rest of the setup that it warrants me to replace the water block with a heatsink?
Let's get back on track and not argue about the other stuff please.
 

iwantanewcomputer

Diamond Member
Apr 4, 2004
5,045
0
0
in general there is no northbridge that makes enough heat to merit watercooling unless you want insane fsb speeds (275+) and are already vmodding the chipset. i doubt you would be able to get any higher fsb speeds by watercooling than by just taking a quality heatsing for a processor and arctic silver epoxying it to the northbridge
 

dmw16

Diamond Member
Nov 12, 2000
7,608
0
0
While engineers are weighing in on this...how about letting the aerospace engineering guy weigh-in :)

I too know a little bit about fluid flows through all sorts of things. :)

Anyway, what the chemical engineer said was true. I just wanted to clarify one thing. The velocity can vary across the system. it is the mass flow rate that remains constant. So for an incompressible working fluid (such as water) mass flow rate (mdot) is equal to density(rho)*area(A)*velocity(V).

so mdot = rho*A*V

so the velocity change will be primarly changed due to changes in cross sectional flow area. Although there will be some back pressure developed around a 90degree bend. I suppose there could be some stagnation point around the bend, but that is far outside the scope of this question.

I think in this case, you should ignore engineering "logic" and listen to the guys w/ real experience water cooling systems.
 

gotensan01

Golden Member
Jul 6, 2004
1,446
0
0
Originally posted by: iwantanewcomputer
in general there is no northbridge that makes enough heat to merit watercooling unless you want insane fsb speeds (275+) and are already vmodding the chipset. i doubt you would be able to get any higher fsb speeds by watercooling than by just taking a quality heatsing for a processor and arctic silver epoxying it to the northbridge
I completely understand that I do not NEED to watercool my chipset.

Here is my dilema. Should I get a better heatsink (stock asus one blows) and take the trouble to remove the chipset water block from my system? Or should I leave it in there because at the least it looks cool? From what I've read thus far, having the chipset block in there does not hinder performance of the entire system enough to remove it.