Overclock.net › Forums › Cooling › Water Cooling › Single vs. dual pass radiators - "Cross-flow"
New Posts  All Forums:Forum Nav:

Single vs. dual pass radiators - "Cross-flow"

post #1 of 35
Thread Starter 
I'll be putting together a water cooling loop within the next two months and have been pondering about this for a week or so now. Basically, I'll be using a bay mounted res/pump and would love to have a cross flow rad because I could easily hide the first bit of tubing inside the bay and then come out of the rad on the other side and go straight into CPU block. However, there's a few problems;

For one, a ton of people really seem to disfavor cross flow radiators for some reason, claiming that they are clearly inferior because they are designed for single pass flow.

Secondly, I could technically use a thicker radiator. Cross flow pretty much limits me to using 30mm thick rad, while my case could fit a 45-50mm. I'm pretty sure I'll be somewhat fine cooling-wise with a push-pull 30mm radiator, but it's very hard to make an estimate without any solid data.

Now I can't really find any reviews or charts online detailing the performance difference of dual vs. single pass flow so could somebody give me advice on this matter?
The bottom line; Does cross flow really suck compared to single flow and if so, why?
post #2 of 35
The only reason why a cross-flow radiator performs slightly under a dual pass radiator is because the waterflow in the tubes is half as fast.
With a dual pass radiator the water goes forth through 6 parallel tubes, and back to 6 other. In other words the two sets of 6 parallel tubes are in series with each other. With crossflow/single pass the water goes through all 12 tubes, which means the surface area the water travels through is twice as big so the velocity of the water is half as high with the same GPM.

The reason faster moving water disposes it's heat better is because when water is turbulent every water-particle directly contacts the copper tubing at one point.
With laminar flow the heat has to travel through the water itself and water just isn't such a good heat conductor as copper or any other metal.
The more turbulent the flow the more efficient the heat transfer will become. However this is not a linear process and at a certain point there's not much to gain. A big difference would be visible at the point where the water turns to a laminar flow but you're well above that point with either radiator since at such a low flow your waterblock wouldn't be able to pick up much heat anyways.


Bottom line, if you want the aesthetics of the crossflow and don't mind a 15mm thinner rad then go for it. If you want aesthetics ánd the 15mm extra rad thickness you could go for an Alphacool rad. They have a fillport in the top so if you connect the reservoir/pump in the drivebay to both bottom ports and use the fillport in the top as exit you've made yourself a crossflow radiator that's 44mm thick (or thicker if you decide to go only push)
post #3 of 35
Quote:
Originally Posted by Jetskyer View Post

The only reason why a cross-flow radiator performs slightly under a dual pass radiator is because the waterflow in the tubes is half as fast.
With a dual pass radiator the water goes forth through 6 parallel tubes, and back to 6 other. In other words the two sets of 6 parallel tubes are in series with each other. With crossflow/single pass the water goes through all 12 tubes, which means the surface area the water travels through is twice as big so the velocity of the water is half as high with the same GPM.

The reason faster moving water disposes it's heat better is because when water is turbulent every water-particle directly contacts the copper tubing at one point.
With laminar flow the heat has to travel through the water itself and water just isn't such a good heat conductor as copper or any other metal.
The more turbulent the flow the more efficient the heat transfer will become. However this is not a linear process and at a certain point there's not much to gain. A big difference would be visible at the point where the water turns to a laminar flow but you're well above that point with either radiator since at such a low flow your waterblock wouldn't be able to pick up much heat anyways.


Bottom line, if you want the aesthetics of the crossflow and don't mind a 15mm thinner rad then go for it. If you want aesthetics ánd the 15mm extra rad thickness you could go for an Alphacool rad. They have a fillport in the top so if you connect the reservoir/pump in the drivebay to both bottom ports and use the fillport in the top as exit you've made yourself a crossflow radiator that's 44mm thick (or thicker if you decide to go only push)


Correct me if I'm wrong, but isn't the ideal cooling situation when you have two fluids exchanging heat to have the fluid BEING cooled to move slowly, and the fluid DOING the cooling moving fast, thus allowing the fluid being cooled to have more time in contact with the fluid that is doing the cooling?

Unless I'm mistaken a normal radiator wouldn't have any more turbulence in the liquid than a xflow rad because, because they both use straight chambers. I must be missing something SOMEWHERE!
Edited by ZytheEKS - 6/1/13 at 6:23pm
The Laboratory
(11 items)
 
  
CPUMotherboardGraphicsRAM
AMD FX-8350 Vishera 8-Core 4.0GHz Asus Crosshair V Formula-Z Radeon R9 290X Generic 1333MHz R.A.M. 
Hard DriveOptical DriveCoolingOS
Wester Digital 5400RPM HP CH20L "The Russian Winter" Windows 8 
KeyboardPowerCase
Steel Series Merc Fatal1ty 1000Watt PSU DIY Rig 
  hide details  
Reply
The Laboratory
(11 items)
 
  
CPUMotherboardGraphicsRAM
AMD FX-8350 Vishera 8-Core 4.0GHz Asus Crosshair V Formula-Z Radeon R9 290X Generic 1333MHz R.A.M. 
Hard DriveOptical DriveCoolingOS
Wester Digital 5400RPM HP CH20L "The Russian Winter" Windows 8 
KeyboardPowerCase
Steel Series Merc Fatal1ty 1000Watt PSU DIY Rig 
  hide details  
Reply
post #4 of 35
Quote:
Originally Posted by ZytheEKS View Post

Correct me if I'm wrong, but isn't the ideal cooling situation when you have two fluids exchanging heat to have the fluid BEING cooled to move slowly, and the fluid DOING the cooling moving fast, thus allowing the fluid being cooled to have more time in contact with the fluid that is doing the cooling?

Unless I'm mistaken a normal radiator wouldn't have any more turbulence in the liquid than a xflow rad because, because they both use straight chambers. I must be missing something SOMEWHERE!

This is not true for a closed loop where a given amount of heat enters (the cpu load). When you have slow moving water it does stay in the radiator longer and will dissipate more heat (though less efficient as I explained in my previous post) however the water will stay longer in the cpu block as well and will be hotter to begin with.
Therefore we try to reach for higher flow since the water is more efficient at absorbing/disposing the heat and we can see the loop as a constant temperature medium so you don't have to worry about loop order as well.
post #5 of 35
There WAS a lot of test data on he subject when water cooling radiators were first developing. Perhaps Bill Adams was the one that did all that work for thermal management. I can not recall the details but I do recall the discussion and dual pass designs generally were the best balance. Look at any C/W chart relative to flow and you will see some benefits of increased flow but it is generally pretty minimal past the 1GPM mark.

I have heard rumors of Quad pass designs in prototype form but nothing I know of gone to mass production. The only thing really different are the HWlabs GTS and GTX series which utilize thinner than normal tubes which provides a similar benefit to quad pass and also allows less air flow restriction from the water tubes.

X flow should be roughly have as restrictive but with most 2-pass rads being fairly low in restriction, It doesn't make sense to go x-flow unless tubing routing is priority over ultimate performance.

I suspect most of that old test data was based on slim thickness designs though so I do think there is merit to quad pass on premium double thickness rads. Does require more coplex plenum chambers though. 3-pass x-flow makes more sense to me.

Anyhow, going off of memory but that's what I've heard. Never did any testing myself.

And NO slow flow rate does not increase performance. The cooling doesn't happen in one cycle, you need to think about time spent in the rad over many many cycles and it is the same regardless of flow rate. It can take upwards of 40min to reach equilibrium on a loop. If you have .2gallons in the loop how long does the fluid stay in the radiator at 1GPm vs .5GPM over that 40 minute period. It is the same. The higher flow rate of 1GPM will always perform better though due to better heat transfer and more turbulent flow.
    
CPUMotherboardGraphicsRAM
3930K ASrock X79 Ext7 Evga GTX570 CORSAIR Vengeance 16GB 
Hard DriveOptical DriveOSMonitor
Phoenix SSD+2TB Seagate drives ASUS DVD-RW combo Windows 7 28" Hanns-G HZ281 
KeyboardPowerCaseMouse
MS Natural Corsair TX650 Danger Den Torture Rack Logitech MX518 
Mouse Pad
Desk 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
3930K ASrock X79 Ext7 Evga GTX570 CORSAIR Vengeance 16GB 
Hard DriveOptical DriveOSMonitor
Phoenix SSD+2TB Seagate drives ASUS DVD-RW combo Windows 7 28" Hanns-G HZ281 
KeyboardPowerCaseMouse
MS Natural Corsair TX650 Danger Den Torture Rack Logitech MX518 
Mouse Pad
Desk 
  hide details  
Reply
post #6 of 35
Quote:
Originally Posted by Jetskyer View Post

This is not true for a closed loop where a given amount of heat enters (the cpu load). When you have slow moving water it does stay in the radiator longer and will dissipate more heat (though less efficient as I explained in my previous post) however the water will stay longer in the cpu block as well and will be hotter to begin with.
Therefore we try to reach for higher flow since the water is more efficient at absorbing/disposing the heat and we can see the loop as a constant temperature medium so you don't have to worry about loop order as well.


I was referring to X-Flow vs U-Pass radiators. X-Flow would decrease the speed of the flowing liquid by dividing it into more channels, thus allowing the liquid to be in contact with the metal chambers/fins, which /should/ result in better cooling, or am I missing something?
The Laboratory
(11 items)
 
  
CPUMotherboardGraphicsRAM
AMD FX-8350 Vishera 8-Core 4.0GHz Asus Crosshair V Formula-Z Radeon R9 290X Generic 1333MHz R.A.M. 
Hard DriveOptical DriveCoolingOS
Wester Digital 5400RPM HP CH20L "The Russian Winter" Windows 8 
KeyboardPowerCase
Steel Series Merc Fatal1ty 1000Watt PSU DIY Rig 
  hide details  
Reply
The Laboratory
(11 items)
 
  
CPUMotherboardGraphicsRAM
AMD FX-8350 Vishera 8-Core 4.0GHz Asus Crosshair V Formula-Z Radeon R9 290X Generic 1333MHz R.A.M. 
Hard DriveOptical DriveCoolingOS
Wester Digital 5400RPM HP CH20L "The Russian Winter" Windows 8 
KeyboardPowerCase
Steel Series Merc Fatal1ty 1000Watt PSU DIY Rig 
  hide details  
Reply
post #7 of 35
Smaller/more channels is better for low-temperature differential cooling as in PC watercooling.
Once again...
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7 920 [4.28GHz, HT] Asus P6T + Broadcom NetXtreme II VisionTek HD5850 [900/1200] + Galaxy GT240 2x4GB G.Skill Ripjaw X [1632 MHz] 
Hard DriveOSMonitorKeyboard
Intel X25-M 160GB + 3xRAID0 500GB 7200.12 Window 7 Pro 64 Acer H243H + Samsung 226BW XARMOR-U9BL  
PowerCaseMouseMouse Pad
Antec Truepower New 750W Li Lian PC-V2100 [10x120mm fans] Logitech G9 X-Trac Pro 
  hide details  
Reply
Once again...
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7 920 [4.28GHz, HT] Asus P6T + Broadcom NetXtreme II VisionTek HD5850 [900/1200] + Galaxy GT240 2x4GB G.Skill Ripjaw X [1632 MHz] 
Hard DriveOSMonitorKeyboard
Intel X25-M 160GB + 3xRAID0 500GB 7200.12 Window 7 Pro 64 Acer H243H + Samsung 226BW XARMOR-U9BL  
PowerCaseMouseMouse Pad
Antec Truepower New 750W Li Lian PC-V2100 [10x120mm fans] Logitech G9 X-Trac Pro 
  hide details  
Reply
post #8 of 35
Quote:
Originally Posted by Martinm210 View Post

There WAS a lot of test data on he subject when water cooling radiators were first developing. Perhaps Bill Adams was the one that did all that work for thermal management. I can not recall the details but I do recall the discussion and dual pass designs generally were the best balance. Look at any C/W chart relative to flow and you will see some benefits of increased flow but it is generally pretty minimal past the 1GPM mark.

I have heard rumors of Quad pass designs in prototype form but nothing I know of gone to mass production. The only thing really different are the HWlabs GTS and GTX series which utilize thinner than normal tubes which provides a similar benefit to quad pass and also allows less air flow restriction from the water tubes.

X flow should be roughly have as restrictive but with most 2-pass rads being fairly low in restriction, It doesn't make sense to go x-flow unless tubing routing is priority over ultimate performance.

I suspect most of that old test data was based on slim thickness designs though so I do think there is merit to quad pass on premium double thickness rads. Does require more coplex plenum chambers though. 3-pass x-flow makes more sense to me.

Anyhow, going off of memory but that's what I've heard. Never did any testing myself.

And NO slow flow rate does not increase performance. The cooling doesn't happen in one cycle, you need to think about time spent in the rad over many many cycles and it is the same regardless of flow rate. It can take upwards of 40min to reach equilibrium on a loop. If you have .2gallons in the loop how long does the fluid stay in the radiator at 1GPm vs .5GPM over that 40 minute period. It is the same. The higher flow rate of 1GPM will always perform better though due to better heat transfer and more turbulent flow.


So take the Black Ice GT Stealth, and the Black ice GT Stealth X-Flow. The liquid is divided into twice as many chambers on a single pass on an X-Flow, so it would be moving slower, would it not? Technically it has the same flowrate, but the velocity of the liquid is slower allowing more time for the air to cool the liquid. I've also heard the argument that less restriction will increase the velocity of the liquid going through it, but the loops speed is ultimately dependent on your waterblocks, as they have the most restriction.
Edited by ZytheEKS - 6/1/13 at 8:03pm
The Laboratory
(11 items)
 
  
CPUMotherboardGraphicsRAM
AMD FX-8350 Vishera 8-Core 4.0GHz Asus Crosshair V Formula-Z Radeon R9 290X Generic 1333MHz R.A.M. 
Hard DriveOptical DriveCoolingOS
Wester Digital 5400RPM HP CH20L "The Russian Winter" Windows 8 
KeyboardPowerCase
Steel Series Merc Fatal1ty 1000Watt PSU DIY Rig 
  hide details  
Reply
The Laboratory
(11 items)
 
  
CPUMotherboardGraphicsRAM
AMD FX-8350 Vishera 8-Core 4.0GHz Asus Crosshair V Formula-Z Radeon R9 290X Generic 1333MHz R.A.M. 
Hard DriveOptical DriveCoolingOS
Wester Digital 5400RPM HP CH20L "The Russian Winter" Windows 8 
KeyboardPowerCase
Steel Series Merc Fatal1ty 1000Watt PSU DIY Rig 
  hide details  
Reply
post #9 of 35
More time for the rad to cool the water is exactly what you don't want. Imagine a great Hypothetical rad that water enters at 10C and is cooled to 5C by the time it gets to the outlet. At the start of the rad the tubes are dissipating 200 watts with 1500 RPM fans and the 10C delta ... by the end of the rad its dissipating half that, 5C delta = 100 watts at 1500RPM ... so the average cooling for the length of the rad is 150 watts.
Now speed the water flow up so that the water leaving the rad is only cooled to 9C ... Now the average for the length of the rad is 195 watts.

You need to think of the cooling loop as a single entity ... heat is created at one one point, the blocks, and given off to the atmosphere at another point, the rads. To keep the entire water volume temp under control we want the best possible heat transfer to the atmosphere and that means keeping the rad as hot as possible.
Kusanagi
(21 items)
 
  
CPUMotherboardGraphicsRAM
Intel Core i5 3570K Asus Sabertooth Z77 EVGA GTX 780 SC Corsair Vengeance Pro CMY16GX3M2A2133C11 16GB 
Hard DriveCoolingCoolingCooling
Samsung 840 Pro 256GB X2 EK Supremacy Aqua Computer Aquaero 6 Pro Aqua Computer D5 USB 
CoolingCoolingCoolingCooling
Aquacomputer MPS 400 flow meter EK-CoolStream PE 360 EK-Coolstream PE 240 Noiseblocker Eloop B12-PS 
CoolingCoolingCoolingCooling
EK Vardar F2 EK-FC Titan Acetal & Nickel EK-RAM Monarch X4 Clean CSQ - Acetal+Nickel EK-D5 X-RES 140 CSQ 
OSMonitorMonitorPower
Win 8.1 Pro 64 bit Samsung S27A750d LG E2290V-SN Silverstone Strider Plus 850W ST85F-P 
Case
Corsair Obsidian 750D 
  hide details  
Reply
Kusanagi
(21 items)
 
  
CPUMotherboardGraphicsRAM
Intel Core i5 3570K Asus Sabertooth Z77 EVGA GTX 780 SC Corsair Vengeance Pro CMY16GX3M2A2133C11 16GB 
Hard DriveCoolingCoolingCooling
Samsung 840 Pro 256GB X2 EK Supremacy Aqua Computer Aquaero 6 Pro Aqua Computer D5 USB 
CoolingCoolingCoolingCooling
Aquacomputer MPS 400 flow meter EK-CoolStream PE 360 EK-Coolstream PE 240 Noiseblocker Eloop B12-PS 
CoolingCoolingCoolingCooling
EK Vardar F2 EK-FC Titan Acetal & Nickel EK-RAM Monarch X4 Clean CSQ - Acetal+Nickel EK-D5 X-RES 140 CSQ 
OSMonitorMonitorPower
Win 8.1 Pro 64 bit Samsung S27A750d LG E2290V-SN Silverstone Strider Plus 850W ST85F-P 
Case
Corsair Obsidian 750D 
  hide details  
Reply
post #10 of 35
Quote:
Originally Posted by ZytheEKS View Post

So take the Black Ice GT Stealth, and the Black ice GT Stealth X-Flow. The liquid is divided into twice as many chambers on a single pass on an X-Flow, so it would be moving slower, would it not? Technically it has the same flowrate, but the velocity of the liquid is slower allowing more time for the air to cool the liquid. I've also heard the argument that less restriction will increase the velocity of the liquid going through it, but the loops speed is ultimately dependent on your waterblocks, as they have the most restriction.

Yes moving slower, but no not in the radiator longer (relative to stability timeframes) and no not better. U flow or two pass is better.

If slower was better, you could take a normal CPU block and mill out the pins or microchannels. This will net you an increased cross sectional area and net you the same slower velocity, but not better temps.

This very common rad misunderstanding comes up all the time. Reason why it is confusing because cooling does NOT happen in one cycle. Time relative to multiple cycles is the same no matter how many passes if the flow rate is the same. Have to stop thinking the cooling happens in one cycle, it does not. one cycle will often only drop temps a fraction of a degree. It's the constant recycling that cools the water down and faster velocity means better thermal performance in general which is why we have two pass designs for the most part.

The ONLY benefit of XFlow is convenience of tube routing if your loop benefits in tube routing needs. It is also less restrictive but relative to CPU block restriction a moot benefit.

Trust the industry, 99% of the radiators are two pass U-Flow designs because it is the optimal balance for most rads. More passes means more restriction without much benefit and possible causing more cpu flow rate loss than benefits in rad. Less is the opposite.

X-flow is like batwings fan blades or red led fans, purely aesthetic and doesn't actually perform better. It actually degrades performance a little on most rad designs. The GTS is however a little different animal. Ultra slim tubes and extreme restriction compared to standard rads so who know on that particular model.
Any real radiator should come with c/w vs flow rate charts to show you. If there is no data then It's not a proven benefit in my book. People have proven two pass to work best for normal rads, nobody has shown otherwise. I do however speculate quad pass may be best for double row premium thickness rads since you have parallel rows it makes sense that quad pass would do beat on double thickness designs.

Have heard rumors of the concept being tested on prototypes but never seen the concept emerge in a mass produced premium double thickness rad.
Edited by Martinm210 - 6/1/13 at 8:45pm
    
CPUMotherboardGraphicsRAM
3930K ASrock X79 Ext7 Evga GTX570 CORSAIR Vengeance 16GB 
Hard DriveOptical DriveOSMonitor
Phoenix SSD+2TB Seagate drives ASUS DVD-RW combo Windows 7 28" Hanns-G HZ281 
KeyboardPowerCaseMouse
MS Natural Corsair TX650 Danger Den Torture Rack Logitech MX518 
Mouse Pad
Desk 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
3930K ASrock X79 Ext7 Evga GTX570 CORSAIR Vengeance 16GB 
Hard DriveOptical DriveOSMonitor
Phoenix SSD+2TB Seagate drives ASUS DVD-RW combo Windows 7 28" Hanns-G HZ281 
KeyboardPowerCaseMouse
MS Natural Corsair TX650 Danger Den Torture Rack Logitech MX518 
Mouse Pad
Desk 
  hide details  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Water Cooling
Overclock.net › Forums › Cooling › Water Cooling › Single vs. dual pass radiators - "Cross-flow"