Overclock.net › Forums › Cooling › Water Cooling › Pipe bending 101
New Posts  All Forums:Forum Nav:

Pipe bending 101 - Page 90

post #891 of 2160
Yeah, the pressure don't worry me any more. They'are just... tiny.
My gaming desk
(14 items)
 
Miner
(15 items)
 
 
CPUMotherboardGraphicsRAM
AMD FX-8350 ASUS Sabertooth 990FX Sapphire 290x Patriot Viper Extreme Division 4 
Hard DriveHard DriveCoolingCooling
OCZ Vertex 460 WD 1TB Blue Koolance 380A EK-FC R9-290X 
OSMonitorKeyboardPower
Windows 8.1 64-bit Monoprice 27" IPS-Glass Pro Panel DUCKY! Antec HCP 850W 
CaseMouse
Hand built wooden desk Logitech G700S 
CPUMotherboardGraphicsGraphics
AMD A10 7850K ASUS A88X-Pro Sapphire 280x Sapphire 280x 
RAMHard DriveHard DriveCooling
Patriot Viper Xtreme Div4 1866mhz 60GB Vertex2 WD Caviar 500AAKS XSPC Rasa 
OSPowerCase
Windows 7 64 bit EVGA 1000W G2 Thermaltake Level 10 GTS 
  hide details  
Reply
My gaming desk
(14 items)
 
Miner
(15 items)
 
 
CPUMotherboardGraphicsRAM
AMD FX-8350 ASUS Sabertooth 990FX Sapphire 290x Patriot Viper Extreme Division 4 
Hard DriveHard DriveCoolingCooling
OCZ Vertex 460 WD 1TB Blue Koolance 380A EK-FC R9-290X 
OSMonitorKeyboardPower
Windows 8.1 64-bit Monoprice 27" IPS-Glass Pro Panel DUCKY! Antec HCP 850W 
CaseMouse
Hand built wooden desk Logitech G700S 
CPUMotherboardGraphicsGraphics
AMD A10 7850K ASUS A88X-Pro Sapphire 280x Sapphire 280x 
RAMHard DriveHard DriveCooling
Patriot Viper Xtreme Div4 1866mhz 60GB Vertex2 WD Caviar 500AAKS XSPC Rasa 
OSPowerCase
Windows 7 64 bit EVGA 1000W G2 Thermaltake Level 10 GTS 
  hide details  
Reply
post #892 of 2160
And you have to remember that pipping wont give you those kinking problems with normal tubing does,
post #893 of 2160
post #894 of 2160
I wrote the following to try and help with some of the theoretical considerations going into pipe flow and assist in loop configs here for people. I bow down to B-negatives knowledge of realworld materials and the plumbing practice (I claim no knowledge in this area and he is awesome!) What I have written more is to address theory and maybe clear up reasoning behind a decision you want to make about a loop, how many bends, sections, etc.

My credentials:
1. Ocean engineer from #1 program in the United states (top 10 engineering school for all other programs)
2. I've had 4 classes directly addressing closed and open fluid flow, and all other classes involved fluids in some way (3.45/4 gpa)
3. Will be working for EMAS AMC as Subsea pipeline systems engineer starting in June 2013

I tell you this not to brag or any such thing, simply to validate the things I have written were done by someone not completely speaking from thin air and no background




Hey guys love the thread but just some insights from an ocean engineer....

1. pipe bends will add to pipe restriction (but not by much). The entrance and exit points for pipe sections add the most restrictions for flow. The math is done with coeffieceints and a bend gets around .15 while and entrance and exit is around a 1 in restriction coefficients math when added up to make pressure calculations. THEREFORE when piping keep the sections/segments (which transaltes to exits/entrances) of pipe to a minimum but realistically make as many bends as you need it won't matter unless changing direction 90 degrees or more, or having an obscene amount of bends. almost 7 bends = one exit/entrance point


2. placing double pipes (tandem) will not offer any increased amount of fluid flow if you use those "y" fittings. the reason for this being that the flow rate Q will be choked down at the consolidation point which is those fittings. if you can have two pressured tubes run from something like say a res to another res then the flow rate will be more or dual pumps into one y fitting, but if you have to consolidate down to one tube or fitting at some point then NO your flow rate or "amount of fluid moved per unit of time" in something like liters per minute or gpm will not be increased at all beyond the flow rate through the single tubed sections.


SO any one who wants to do this for aesthetics, rock on it will look sick! thumb.gif BUT you will get ZERO performance upgrade or higher flow to components because obviously to flow through a pc block there's only one channel right? gpu blocks are the same, so unless its a multiple input/ output block with multiple channels then you will see no gains and will see resistance (pump head) increase due to multiple restrictions introduced into the loop

the only way this will help is because there is more fluid in the system, which will give you more thermal overhead initially but once all the fluid is heated up to the same levels it wont matter much and the difference might become negligible (this is why WC has not evolved to include massive reservoirs, once the system is at a thermal equilibrium it doesn't matter how much fluid is in it, its closed, the only thing taking away heat is the dissipation in the radiators)

in addition when thinking about larger diameter tubes anticipating more flow. The flow you will get in your loop will never exceed the pumps movement amounts (which i figure everyone knows) but to clarify larger od will move fluid slower than smaller od in relation to the od for the in/out at the pumps.

ie: if the pump has 10 gpm (made up number) with od out of 1/2" then you increase the od of the pipe in the loop to 3/4" the flow rate at any point in the 3/4 tube will not be the same speed as leaving the 10 gpm pump and for every point that the flow profile diameter (pipe diameter gets larger or smaller it will cause restriction to flow (no matter bigger or smaller) think of it like getting smaller the pump must work harder to move same amount of water through the pipe and larger causes the pipe to require more fluid volume per inch to fill and causes more work for the pump at the interface change.

let me be clear IF the pump is rated at 10 gpm, then it will deliver 10 gpm....but it is only moving that parcel of water and nothing more or less if set at that speed. all you are doing with larger od is keeping the speed at a given point in a loop slower because the fluid parcel is spread over a larger diameter as it moves (per unit force being put out by the pump its only going to move so much volume so far and larger od makes the volume per inch larger and more energy needed to move the same distance hence slower.... smaller od is like a rocket engine it will move faster through a pipe due to the same principle in reverse.

roughness factor of copper vs. acrylic vs. stainless etc. should not really matter for loops this small in a pc to where it really would govern the pipe choice (most pc WC pumps are overkill and won't be affected by a slightly more ore less rough pipe material

AND difference between smaller and larger diameter pipes why or why not? if you have a larger spread of a loop where you need speed to keep momentum to make it to everything and lots of restrictions id say go smaller.....if you have a really heavy duty loop (lots of things needing heat disipation) then I'd say go for more OD because the slower moving loop will absorb more heat as it moves and take it away from the components (this should be done withing reason though, too slow is obviously bad as it wont dissipate fast enough.

WC is 100% about a balance between contact surface area of water per a unit of time ( how long it is in contact) which is directly related to the speed of a loop and amount of water moved. So a super fast loop will not do much for you, or a super slow loop will not do well either. variable pumps will help with this balance but another way of slowing or speeding a loop involves the choice of OD of your pipes.

VERTICAL distances also matter, I won't go into it but hydro-statically (weight of water) this translates directly into pump head which is what will affect your loop the most as it is what the pump is fighting to overcome.

Is this important at all? depends

the small variations in OD relatively (3/8-3/4") and small variations of loop length ( 6ft-10ft) these concepts obviously apply and they are scalable (they work the same at massive pipe levels or small. but given the over engineering that goes into enthusiast WC pumps most use on OCN....it may or may not be of interest to make all these considerations. I'd say realistically it may only be the difference between 3-5C or more Conservatively 2-3C if you make the right optimizations. But for anyone looking to push machines to the absolute limits or find that balance of fan noise they making these choices could make the difference for you...

I wrote this to try and HELP! not to hurt any ones feelings or degrade your decisions! I think everyone's work here is great and I hope to join the club sometime this year when i get a chance and the time. Knowledge is power so I'm just trying to spread the love a bit and see if it helps anyone here.

If you read the whole thing then cheers! and hope you found it worth the read and relevant thumb.gifbiggrin.gif
Edited by oats2012 - 3/31/13 at 2:15am
Gaming PORTAL
(19 items)
 
  
CPUMotherboardGraphicsRAM
3770k msi z77 mpower PowerColor 7870 (Tahiti LE) EZ G.SKILL Sniper Series 16GB (4 x 4GB) 240-Pin DD... 
Hard DriveHard DriveHard DriveOptical Drive
Intel 320 Series SSDSA2CW120G3K5 2.5" 120GB SAT... Samsung 840 WD caviar black ASUS 24X DVD Burner - Bulk 
CoolingOSMonitorKeyboard
NH-D14 Windows 7 ultimate 3 x 23" monitors Logitech G110 
PowerCaseMouseMouse Pad
COOLER MASTER Silent Pro Gold Series RS800-80GA... Corsair C70 Logitech G500 Mionix Sargas 320 
AudioAudioAudio
Pioneer VSX-821-K Sony DH500 polk Monitor II 40's & Monitor II 70's 
  hide details  
Reply
Gaming PORTAL
(19 items)
 
  
CPUMotherboardGraphicsRAM
3770k msi z77 mpower PowerColor 7870 (Tahiti LE) EZ G.SKILL Sniper Series 16GB (4 x 4GB) 240-Pin DD... 
Hard DriveHard DriveHard DriveOptical Drive
Intel 320 Series SSDSA2CW120G3K5 2.5" 120GB SAT... Samsung 840 WD caviar black ASUS 24X DVD Burner - Bulk 
CoolingOSMonitorKeyboard
NH-D14 Windows 7 ultimate 3 x 23" monitors Logitech G110 
PowerCaseMouseMouse Pad
COOLER MASTER Silent Pro Gold Series RS800-80GA... Corsair C70 Logitech G500 Mionix Sargas 320 
AudioAudioAudio
Pioneer VSX-821-K Sony DH500 polk Monitor II 40's & Monitor II 70's 
  hide details  
Reply
post #895 of 2160
Thread Starter 
Quote:
Originally Posted by oats2012 View Post

I wrote the following to try and help with some of the theoretical considerations going into pipe flow and assist in loop configs here for people. I bow down to B-negatives knowledge of realworld materials and the plumbing practice (I claim no knowledge in this area and he is awesome!) What I have written more is to address theory and maybe clear up reasoning behind a decision you want to make about a loop, how many bends, sections, etc.

My credentials:
1. Ocean engineer from #1 program in the United states (top 10 engineering school for all other programs)
2. I've had 4 classes directly addressing closed and open fluid flow (3.4/4 gpa)
3. Will be working for EMAS AMC as Subsea pipeline systems engineer starting in June 2013

I tell you this not to brag or any such thing, simply to validate the things I have written were done by someone not completely speaking from thin air and no background




Hey guys love the thread but just some insights from an ocean engineer....

1. pipe bends will add to pipe restriction (but not by much). The entrance and exit points for pipe sections add the most restrictions for flow. The math is done with coeffieceints and a bend gets around .15 while and entrance and exit is around a 1 in restriction coefficients math when added up to make pressure calculations. THEREFORE when piping keep the sections/segments (which transaltes to exits/entrances) of pipe to a minimum but realistically make as many bends as you need it won't matter unless changing direction 90 degrees or more, or having an obscene amount of bends. almost 7 bends = one exit/entrance point


2. placing double pipes (tandem) will not offer any increased amount of fluid flow if you use those "y" fittings. the reason for this being that the flow rate Q will be choked down at the consolidation point which is those fittings. if you can have two pressured tubes run from something like say a res to another res then the flow rate will be more or dual pumps into one y fitting, but if you have to consolidate down to one tube or fitting at some point then NO your flow rate or "amount of fluid moved per unit of time" in something like liters per minute or gpm will not be increased at all beyond the flow rate through the single tubed sections.


SO any one who wants to do this for aesthetics, rock on it will look sick! thumb.gif BUT you will get ZERO performance upgrade or higher flow to components because obviously to flow through a pc block there's only one channel right? gpu blocks are the same, so unless its a multiple input/ output block with multiple channels then you will see no gains and will see resistance (pump head) increase due to multiple restrictions introduced into the loop

in addition when thinking about larger diameter tubes anticipating more flow. The flow you will get in your loop will never exceed the pumps movement amounts (which i figure everyone knows) but to clarify larger od will move fluid slower than smaller od in relation to the od for the in/out at the pumps.

ie: if the pump has 10 gpm (made up number) with od out of 1/2" then you increase the od of the pipe in the loop to 3/4" the flow rate at any point in the 3/4 tube will not be the same speed as leaving the 10 gpm pump and for every point that the flow profile diameter (pipe diameter gets larger or smaller it will cause restriction to flow (no matter bigger or smaller) think of it like getting smaller the pump must work harder to move same amount of water through the pipe and larger causes the pipe to require more fluid volume per inch to fill and causes more work for the pump at the interface change.

let me be clear IF the pump is rated at 10 gpm, then it will deliver 10 gpm....but it is only moving that parcel of water and nothing more or less if set at that speed. all you are doing with larger od is keeping the speed at a given point in a loop slower because the fluid parcel is spread over a larger diameter as it moves (per unit force being put out by the pump its only going to move so much volume so far and larger od makes the volume per inch larger and more energy needed to move the same distance hence slower.... smaller od is like a rocket engine it will move faster through a pipe due to the same principle in reverse.

roughness factor of copper vs. acrylic vs. stainless etc. should not really matter for loops this small in a pc to where it really would govern the pipe choice (most pc WC pumps are overkill and won't be affected by a slightly more ore less rough pipe material

AND difference between smaller and larger diameter pipes why or why not? if you have a larger spread of a loop where you need speed to keep momentum to make it to everything and lots of restrictions id say go smaller.....if you have a really heavy duty loop (lots of things needing heat disipation) then I'd say go for more OD because the slower moving loop will absorb more heat as it moves and take it away from the components (this should be done withing reason though, too slow is obviously bad as it wont dissipate fast enough.

Very true,you can only flow volumetrically as much as your narrowest restriction.

The math as to why...http://en.wikipedia.org/wiki/Continuity_equation#Fluid_dynamics

There is a difference in the performance of a 'knuckle' 90 found in most fittings and a 90 tube bend,the 90 bends in tube dont add any real restriction but the knuckles do.
Remember all this example has to do is exhaust a gas,When fitting boiler flues,every knuckle 90 adds a 1m to the tube length...when you only have 3-4 meters to play with anyway. Sweeping bends work with inertia,knuckles dont
Edited by B NEGATIVE - 3/31/13 at 2:25am
MILSPEC II
(8 items)
 
| LUMO |
(6 items)
 
CLoS3 IMPACT
(8 items)
 
CPUMotherboardGraphicsGraphics
5820k X99m WS 980ti strix 980ti strix 
RAMHard DrivePowerCase
KLEVV Cras DDR4  3tb Red Corsair 600w SFX + 450w SFX Caselabs BH4 
CPUMotherboardGraphicsRAM
4970k Z97 Gryphon GTX 680 x2 Crucial Ballistix Elite 16Gb 
PowerCase
Be Quiet Dark Power 850 InWin D Frame Mini 
CPUMotherboardGraphicsRAM
i7 4770k ASUS Impact ITX R9 290 Gskill Trident 2400 
Hard DriveHard DriveCoolingCase
Samsung EV0 250 x2 WD black Loop.....oh yes...... Caselabs S3....modded 
  hide details  
Reply
MILSPEC II
(8 items)
 
| LUMO |
(6 items)
 
CLoS3 IMPACT
(8 items)
 
CPUMotherboardGraphicsGraphics
5820k X99m WS 980ti strix 980ti strix 
RAMHard DrivePowerCase
KLEVV Cras DDR4  3tb Red Corsair 600w SFX + 450w SFX Caselabs BH4 
CPUMotherboardGraphicsRAM
4970k Z97 Gryphon GTX 680 x2 Crucial Ballistix Elite 16Gb 
PowerCase
Be Quiet Dark Power 850 InWin D Frame Mini 
CPUMotherboardGraphicsRAM
i7 4770k ASUS Impact ITX R9 290 Gskill Trident 2400 
Hard DriveHard DriveCoolingCase
Samsung EV0 250 x2 WD black Loop.....oh yes...... Caselabs S3....modded 
  hide details  
Reply
post #896 of 2160
Not sure I follow why water moving slower would necessarily pick up more heat. I thought the contact area of the blocks would have more to do with heat dissipation.
I can certainly understand heat saturation, (water moving too slow) and I understand beyond a point more flow is not going to yield better cooling.
But is it really possible that higher flow can fail to absorb heat as efficiently as a lower flow?

I have more flow than I need, simply for redundancy, and I certainly have overkill on radiator, but I don't see higher flow rates impacting heat dissipation negatively in my loop.

No argument here, enlighten me.
post #897 of 2160
Thread Starter 
Higher flow doesnt impact negatively,it just ceases to increase significantly.
MILSPEC II
(8 items)
 
| LUMO |
(6 items)
 
CLoS3 IMPACT
(8 items)
 
CPUMotherboardGraphicsGraphics
5820k X99m WS 980ti strix 980ti strix 
RAMHard DrivePowerCase
KLEVV Cras DDR4  3tb Red Corsair 600w SFX + 450w SFX Caselabs BH4 
CPUMotherboardGraphicsRAM
4970k Z97 Gryphon GTX 680 x2 Crucial Ballistix Elite 16Gb 
PowerCase
Be Quiet Dark Power 850 InWin D Frame Mini 
CPUMotherboardGraphicsRAM
i7 4770k ASUS Impact ITX R9 290 Gskill Trident 2400 
Hard DriveHard DriveCoolingCase
Samsung EV0 250 x2 WD black Loop.....oh yes...... Caselabs S3....modded 
  hide details  
Reply
MILSPEC II
(8 items)
 
| LUMO |
(6 items)
 
CLoS3 IMPACT
(8 items)
 
CPUMotherboardGraphicsGraphics
5820k X99m WS 980ti strix 980ti strix 
RAMHard DrivePowerCase
KLEVV Cras DDR4  3tb Red Corsair 600w SFX + 450w SFX Caselabs BH4 
CPUMotherboardGraphicsRAM
4970k Z97 Gryphon GTX 680 x2 Crucial Ballistix Elite 16Gb 
PowerCase
Be Quiet Dark Power 850 InWin D Frame Mini 
CPUMotherboardGraphicsRAM
i7 4770k ASUS Impact ITX R9 290 Gskill Trident 2400 
Hard DriveHard DriveCoolingCase
Samsung EV0 250 x2 WD black Loop.....oh yes...... Caselabs S3....modded 
  hide details  
Reply
post #898 of 2160
It's why I enjoy my variable pump.
My gaming desk
(14 items)
 
Miner
(15 items)
 
 
CPUMotherboardGraphicsRAM
AMD FX-8350 ASUS Sabertooth 990FX Sapphire 290x Patriot Viper Extreme Division 4 
Hard DriveHard DriveCoolingCooling
OCZ Vertex 460 WD 1TB Blue Koolance 380A EK-FC R9-290X 
OSMonitorKeyboardPower
Windows 8.1 64-bit Monoprice 27" IPS-Glass Pro Panel DUCKY! Antec HCP 850W 
CaseMouse
Hand built wooden desk Logitech G700S 
CPUMotherboardGraphicsGraphics
AMD A10 7850K ASUS A88X-Pro Sapphire 280x Sapphire 280x 
RAMHard DriveHard DriveCooling
Patriot Viper Xtreme Div4 1866mhz 60GB Vertex2 WD Caviar 500AAKS XSPC Rasa 
OSPowerCase
Windows 7 64 bit EVGA 1000W G2 Thermaltake Level 10 GTS 
  hide details  
Reply
My gaming desk
(14 items)
 
Miner
(15 items)
 
 
CPUMotherboardGraphicsRAM
AMD FX-8350 ASUS Sabertooth 990FX Sapphire 290x Patriot Viper Extreme Division 4 
Hard DriveHard DriveCoolingCooling
OCZ Vertex 460 WD 1TB Blue Koolance 380A EK-FC R9-290X 
OSMonitorKeyboardPower
Windows 8.1 64-bit Monoprice 27" IPS-Glass Pro Panel DUCKY! Antec HCP 850W 
CaseMouse
Hand built wooden desk Logitech G700S 
CPUMotherboardGraphicsGraphics
AMD A10 7850K ASUS A88X-Pro Sapphire 280x Sapphire 280x 
RAMHard DriveHard DriveCooling
Patriot Viper Xtreme Div4 1866mhz 60GB Vertex2 WD Caviar 500AAKS XSPC Rasa 
OSPowerCase
Windows 7 64 bit EVGA 1000W G2 Thermaltake Level 10 GTS 
  hide details  
Reply
post #899 of 2160
I think higher flow is desirable. I don't think that the flow can ever be too fast. But the problem is the pumps. As the flow increases, you also increase the pump's heat output and that heat gets dumped into the loop, especially with a D5 pump.
And soon, the small performance increase from the flow will be voided by the pump's heat dump.

But if you could somehow have no heat dump from the pump, then I don't think the flow would ever be too much.

Think of it this way, a cold day will feel a lot colder if it's windy. Increase the wind (air flow) and you increase the heat absorbtion of the air. Same with water.
Edited by PepeLapiu - 3/31/13 at 9:22am
Dark Vader
(6 items)
 
  
CoolingCoolingCoolingCooling
39 X AP-45 2150 rpm Gentle Typhoon fans Lamptron CW611 FrozenQ Dual Bay Reservoir 2x XT45 480. 1x UT60 480 ~ 1x UT60 360 ~ 1x UT... 
OSCase
Linux Ubuntu CaseLabs STH10 
  hide details  
Reply
Dark Vader
(6 items)
 
  
CoolingCoolingCoolingCooling
39 X AP-45 2150 rpm Gentle Typhoon fans Lamptron CW611 FrozenQ Dual Bay Reservoir 2x XT45 480. 1x UT60 480 ~ 1x UT60 360 ~ 1x UT... 
OSCase
Linux Ubuntu CaseLabs STH10 
  hide details  
Reply
post #900 of 2160
I recently bought the black nickel aquatuning black nickel fittings. I was wondering if these fittings will work with the Acrylic Tubing?
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Water Cooling
Overclock.net › Forums › Cooling › Water Cooling › Pipe bending 101