[O-CuK]Marci
New member
This is another one of those "Myth's of Watercooling" articles... basically dispelling the myth that singlepass is best when it comes to radiators...
Here's the best chunk of txt around that describes why singlepass is generally beaten by dualpass in today's rads, and also why the condenser-style rads such as the
AquaComputer EVO range are only suitable in specific situations.
Now, bear in mind that both the XSPC / AlphaCool / Radiical rads are all identical - only the endtanks are changed to make dualpass into singlepass and vice versa. Same goes for Black Ice. A Black Ice Pro X-Flow is a Black Ice Pro with endtanks swapped - no other change....
The following is the consolidation of a chunk of posts, whipped from [H], and reordered into relevant bits so that singlepass convo is all together... posted by Cathar... (source - http://hardforum.com/showthread.php?t=935607)
Pictures have been inserted for relevance. If it's in italics, I've added it...
=============================================
Single-pass has its drawbacks. It's not appropriate in all situations. The PA160.1 was single pass because it is also single-row. The issue with single pass is the per-tubing velocity which affects water-metal convectional efficiency. If you take what was a dual-pass radiator and blindly make it single pass (ie: Black Ice X-Flow and XSPC examples above - Marci) it can result in a performance loss at low-moderate flow rates especially if the radiator is multiple row (which both XSPC & BI are - Marci)
Okay, a row of tubes, being a parallel row of tubes from one side of the radiator to the other. Imagine it as being like a slice of bread, with tubes running from one end to the other. A single-row radiator is like a single slice of bread. A dual-row radiator is like two slices of bread together.
A few dual-row dual-pass rads:
ThermoChill PA120 Series
Black Ice Pro
XSPC R120-S / Radiical® Pro 120 mm / Alphacool NexXxoS Xtreme I
In a single-pass (cross-flow) radiator the water flows from the inlet at one end, once straight through all the tubes, in all the slices, and to the outlet in the diagonally opposite corner to the inlet.
In a dual-pass radiator the water flows down one half-side of all slices, U-turns, and back up the other half-side of all the slices.
Single-pass benefits because it presents all the tubes with the highest possible water temperature at once, whereas a two-pass radiator will only get the highest temperature inlet water on one side, and then cool slightly cooler water up the other side. This doesn't make a huge difference though. Generally it provides a 1-15% performance benefit from this effect alone.
A designed-to-be single pass single row rad:
ThermoChill PA160
A few dual pass to single pass direct conversions (theoretically bad):
BlackIce Pro X-Flow (ie: Same rad as dual pass above but with endtanks swapped only)
XSPC R120S Crossflow / Radiical® Pro 120 mm Single Fan Radiator- Single Pass (ie: Same rad as dual pass above but with endtanks swapped only)
Where single pass falls down though is the tubing velocity. Because the water is presented to all the tubes at once, the water velocity through the tubes is half of what it is through a dual-pass radiator. Increased water-velocity increases water turbulence, which in turn increases the ability of the water to pass the heat that's stored within it into the metal walls of the tubes that its flowing through.
If you now suddenly halve the water velocity by making a dual-pass radiator into a single pass radiator, you will lose the heat transfer benefit being provided by the higher water velocity through the tubes.
This effect can be seen in terms of radiator heat dissipation based upon changing just the water flow rate alone. eg.
As the flow rate (and hence water velocity in the tubes) goes down, the radiator performance starts to fall away. With a single-pass you've gone and halved the water velocity in one hit. This is offset somewhat by the temperature delta benefit of single-pass, but it is by no means a sure thing that single pass will be better.
Looking at that graph, if we made the HE120.1 into a single pass radiator and our flow rate was 5LPM, then by halving the water velocity with some moderate fans we can see that we'll lose up to 20% convectional efficiency performance. Sure, we gained a bit by going single pass, but we lost a lot due to the per-tube velocity drop.
If our flow rate was 12lpm though, then we only lose ~10%, and maybe (just maybe) single pass might be a better option. Don't know too many people with 12lpm running through their loops though. (Common flow rate in a D5 based system is 6lpm ballpark - Marci) The HE120.1 is a dual-core radiator too, so this example highlights how turning the HE120.1 into a single pass radiator would quite likely be a bad idea.
Now that is just a single (simplistic) example to express what I mean. Ideally such graphs need to be generated for each radiator to make predictions on what's going to happen by going single-pass, but as you can see, it's not a "sure thing". Blindly going single-pass is much like BillA said: people who blindly go high-flow without understanding what they're really trying to achieve. A misguided marketing goal here will drive radiator performance backwards.
The problem here is costs. Heater-core radiators are cheap because they leverage off commonly available components from the car industry. Once you start doing things like custom tube sizes and other custom tweaks that move away from what's commonly available, costs will go up quite quickly, and to be honest, there's probably not a lot to gain in terms of performance. Heater-core style radiators are very efficient items when working well. It tends to more be a case of trying not to lose performance while you're mucking about with building custom ones.
Have to remember that the water-cooling market, in terms of radiators, enjoys its privilege almost solely to the car market. The water-cooling market would be totally different if there were not as much commonality between cars and computers for heat transfer.
The Aqua Computer Evo radiators are coiled copper tubing designs. They are used most frequently in conditions that require high pressures, such as refrigeration condensors. While the coiled tube is great for operating well under low-flow conditions, it is bulky and consumes up a lot of the body of the finned area in comparison to a heater-core style radiator. Additionally, the spacing between the tubes and the fins in the coiled tube radiators makes for a less than optimal setup with respect to the conduction distance.
I suppose a good analogy is that while the looping tube style radiators are good for liquid-metal heat transfer, they are generally far less efficient at metal-air heat transfer when given the same space and size constraints. Heater-core style radiators come close to as good as it gets for metal-air efficiency, and provided you give them sufficient flow rate for their internal design (single pass, multi-pass, whatever) the liquid-metal efficiency can be as good as the looping tube style radiators. Heater-cores fall down though more quickly at lower flow rates than the looping tube style radiators. Given the historically low flow rates that occur in Teutonic systems, the looping tube style radiators make a fair degree of sense.
It's a case of horses for courses. Looping tube can work better at very-low-flow rates, while heater-cores (and heatercore style rads - Marci) tend to work better at low-moderate flow rates and higher.
=============================
Questions from Top Nurse. Rest of txt by Cathar.
There is proper comparative testing data of all the common rads, excluding the PA Series as it wasn't around at the time, by BillA (most noted PC-WaterCooling Radiator Tester worldwide at the moment)... however, this was done whilst Bill was contracted to Swiftech, and once he left Swiftech the data remained their property to do with as they wish and as of yet they have decided not to publish. When and if they do, I'll let you know. The data basically confirms all of the above with facts, figures n graphs.
Here's the best chunk of txt around that describes why singlepass is generally beaten by dualpass in today's rads, and also why the condenser-style rads such as the
AquaComputer EVO range are only suitable in specific situations.
Now, bear in mind that both the XSPC / AlphaCool / Radiical rads are all identical - only the endtanks are changed to make dualpass into singlepass and vice versa. Same goes for Black Ice. A Black Ice Pro X-Flow is a Black Ice Pro with endtanks swapped - no other change....
The following is the consolidation of a chunk of posts, whipped from [H], and reordered into relevant bits so that singlepass convo is all together... posted by Cathar... (source - http://hardforum.com/showthread.php?t=935607)
Pictures have been inserted for relevance. If it's in italics, I've added it...
=============================================
Single-pass has its drawbacks. It's not appropriate in all situations. The PA160.1 was single pass because it is also single-row. The issue with single pass is the per-tubing velocity which affects water-metal convectional efficiency. If you take what was a dual-pass radiator and blindly make it single pass (ie: Black Ice X-Flow and XSPC examples above - Marci) it can result in a performance loss at low-moderate flow rates especially if the radiator is multiple row (which both XSPC & BI are - Marci)
Okay, a row of tubes, being a parallel row of tubes from one side of the radiator to the other. Imagine it as being like a slice of bread, with tubes running from one end to the other. A single-row radiator is like a single slice of bread. A dual-row radiator is like two slices of bread together.
A few dual-row dual-pass rads:
ThermoChill PA120 Series
Black Ice Pro
XSPC R120-S / Radiical® Pro 120 mm / Alphacool NexXxoS Xtreme I
In a single-pass (cross-flow) radiator the water flows from the inlet at one end, once straight through all the tubes, in all the slices, and to the outlet in the diagonally opposite corner to the inlet.
In a dual-pass radiator the water flows down one half-side of all slices, U-turns, and back up the other half-side of all the slices.
Single-pass benefits because it presents all the tubes with the highest possible water temperature at once, whereas a two-pass radiator will only get the highest temperature inlet water on one side, and then cool slightly cooler water up the other side. This doesn't make a huge difference though. Generally it provides a 1-15% performance benefit from this effect alone.
A designed-to-be single pass single row rad:
ThermoChill PA160
A few dual pass to single pass direct conversions (theoretically bad):
BlackIce Pro X-Flow (ie: Same rad as dual pass above but with endtanks swapped only)
XSPC R120S Crossflow / Radiical® Pro 120 mm Single Fan Radiator- Single Pass (ie: Same rad as dual pass above but with endtanks swapped only)
Where single pass falls down though is the tubing velocity. Because the water is presented to all the tubes at once, the water velocity through the tubes is half of what it is through a dual-pass radiator. Increased water-velocity increases water turbulence, which in turn increases the ability of the water to pass the heat that's stored within it into the metal walls of the tubes that its flowing through.
If you now suddenly halve the water velocity by making a dual-pass radiator into a single pass radiator, you will lose the heat transfer benefit being provided by the higher water velocity through the tubes.
This effect can be seen in terms of radiator heat dissipation based upon changing just the water flow rate alone. eg.
As the flow rate (and hence water velocity in the tubes) goes down, the radiator performance starts to fall away. With a single-pass you've gone and halved the water velocity in one hit. This is offset somewhat by the temperature delta benefit of single-pass, but it is by no means a sure thing that single pass will be better.
Looking at that graph, if we made the HE120.1 into a single pass radiator and our flow rate was 5LPM, then by halving the water velocity with some moderate fans we can see that we'll lose up to 20% convectional efficiency performance. Sure, we gained a bit by going single pass, but we lost a lot due to the per-tube velocity drop.
If our flow rate was 12lpm though, then we only lose ~10%, and maybe (just maybe) single pass might be a better option. Don't know too many people with 12lpm running through their loops though. (Common flow rate in a D5 based system is 6lpm ballpark - Marci) The HE120.1 is a dual-core radiator too, so this example highlights how turning the HE120.1 into a single pass radiator would quite likely be a bad idea.
Now that is just a single (simplistic) example to express what I mean. Ideally such graphs need to be generated for each radiator to make predictions on what's going to happen by going single-pass, but as you can see, it's not a "sure thing". Blindly going single-pass is much like BillA said: people who blindly go high-flow without understanding what they're really trying to achieve. A misguided marketing goal here will drive radiator performance backwards.
The problem here is costs. Heater-core radiators are cheap because they leverage off commonly available components from the car industry. Once you start doing things like custom tube sizes and other custom tweaks that move away from what's commonly available, costs will go up quite quickly, and to be honest, there's probably not a lot to gain in terms of performance. Heater-core style radiators are very efficient items when working well. It tends to more be a case of trying not to lose performance while you're mucking about with building custom ones.
Have to remember that the water-cooling market, in terms of radiators, enjoys its privilege almost solely to the car market. The water-cooling market would be totally different if there were not as much commonality between cars and computers for heat transfer.
The Aqua Computer Evo radiators are coiled copper tubing designs. They are used most frequently in conditions that require high pressures, such as refrigeration condensors. While the coiled tube is great for operating well under low-flow conditions, it is bulky and consumes up a lot of the body of the finned area in comparison to a heater-core style radiator. Additionally, the spacing between the tubes and the fins in the coiled tube radiators makes for a less than optimal setup with respect to the conduction distance.
I suppose a good analogy is that while the looping tube style radiators are good for liquid-metal heat transfer, they are generally far less efficient at metal-air heat transfer when given the same space and size constraints. Heater-core style radiators come close to as good as it gets for metal-air efficiency, and provided you give them sufficient flow rate for their internal design (single pass, multi-pass, whatever) the liquid-metal efficiency can be as good as the looping tube style radiators. Heater-cores fall down though more quickly at lower flow rates than the looping tube style radiators. Given the historically low flow rates that occur in Teutonic systems, the looping tube style radiators make a fair degree of sense.
It's a case of horses for courses. Looping tube can work better at very-low-flow rates, while heater-cores (and heatercore style rads - Marci) tend to work better at low-moderate flow rates and higher.
=============================
Questions from Top Nurse. Rest of txt by Cathar.
There is proper comparative testing data of all the common rads, excluding the PA Series as it wasn't around at the time, by BillA (most noted PC-WaterCooling Radiator Tester worldwide at the moment)... however, this was done whilst Bill was contracted to Swiftech, and once he left Swiftech the data remained their property to do with as they wish and as of yet they have decided not to publish. When and if they do, I'll let you know. The data basically confirms all of the above with facts, figures n graphs.
