Loop Order Question...

[MaXXimus]

I was told my loop order should be: Res>Pump>Rad>CPU>Res and I wanted to do: Res>Pump>CPU>Rad>Res. Also was mentioned that I want the coolest liquid hitting my CPU, but wouldnt the loop I want also give the coolest liquid, since it does not hit any heat sources? The reaosn I wanted to do this was because I read that you should use the least amount of tubing as possible and also I read that loop order means nothing as long as you make sure your Res is just before your pump... anyone care to comment?

 

[MaXXimus]

Nobody? Not even a speculation?

 

[silent-circuit]

Doesn't matter either way. Unless you talking about CPU/GPU mixed loops order means nothing.

 

[MrWizard6600]

loop order doesn't really matter so long as your not pumping into the res (obviously). The more tubing you use, the more resistance the water encounters, the less force (pressure) the water will put on your block.

 

[vsboxerboy]

People have differing points of view, but I am of the belief that loop order is not important as your entire loop will eventually reach an equilibrium and no one part of the loop will be significantly hotter than the other.

You are correct it is the more important to use the minimum amount of tubing with as few sharp bends as possible and that it is the most important just to put the res in front of the pump.

 

[nissanztt90]

The system will eventually equalize no matter what. Have the pump draw from the res, nothing else will really matter.

 

[tesseracter]

all this political correctness.

from what ive seen, aiming the pump straight at the waterblock leads to the greatest heat transfer, which is the job of the waterblock. going thru a radiator first eliminates all the good turbulent flow that you want when hitting the CPU. radiator should be moved thru as slow as possible, giving a maximum chance for the water to get as close to ambient temps as possible

i guess im a disciple of Cathar in my belief of Pump->CPU block->Rad->Res->Pump

gets more complicated with GPU blocks, but I would still go pump->CPU and res->pump all the time.

 

[Erasmus354]

Radiators actually dissipate more heat the faster the water flows, not slower. No matter what the flow rate the water will still spend the same amount of time (proportionally to other components) in the radiator. A faster flow rate equates to better thermal transfer between the water and the radiator tubes.

Sure if you think in terms of absolutes, with a faster flow rate the water doesn't cool down as much, but guess what? The water also doesn't heat up as much when going through the heat producing components. When all is said and done, more flow is always better for every single component in a water cooling loop (negating pump heat dump).

 

[tesseracter]

Erasmus, my current setup has a Y branching off after my CPU block into 2 radiators. CPU gets max speed, then the water moves slower in the radiators. this is better for flow than running the same amount of radiator material in series.

remember the little river/cascade thing? worked pretty well, but was a large flow inhibitor. nonetheless, it cooled better at a low flow rate because of inpingement than other blocks at high flow. while it did work better at higher flow rates, the pump needed to work harder, and suddenly you arrived at diminishing returns.

just trying to show you that absolutes are tricky, because of the multiple aspects of flow vs heat transfer. you cannot negate pump heat just because you feel like it, it can make a 2-5C difference in equalized temperature.

 

[nissanztt90]

Its energy in vs energy out. Its simple physics. Given the same variables, everything will equalize. The only variables that matter here are really ambient temp and airflow(fan speed if you will). Man you guys really like to argue you the finer points for an extra 2 degrees dontcha...my time would be better spent chasing a unicorn.

 

[tesseracter]

Its energy in vs energy out. Its simple physics. Given the same variables, everything will equalize. The only variables that matter here are really ambient temp and airflow(fan speed if you will). Man you guys really like to argue you the finer points for an extra 2 degrees dontcha...my time would be better spent chasing a unicorn.

jet impingement, turbulence, fluid flows, and heat conduction are not simple physics. sorry.

given the same variables? /facepalm. we are talking about changing variables.(component order) does that mean i can totally invalidate your post right now?

yes, equalization will happen, but it will happen at different(temps|fan speeds|flow rates) depending on efficiency, which to neglected to mention. a CPU will more efficiently conduct heat to a waterblock if it has themal paste, and a waterblock will more efficiently conduct heat to water if there is a higher flow, AND more turbulence. you wouldnt say that not having a thermal conductor between your CPU and block doesnt matter, would you?

 

[nissanztt90]

jet impingement, turbulence, fluid flows, and heat conduction are not simple physics. sorry.

given the same variables? /facepalm. we are talking about changing variables.(component order) does that mean i can totally invalidate your post right now?

yes, equalization will happen, but it will happen at different(temps|fan speeds|flow rates) depending on efficiency, which to neglected to mention. a CPU will more efficiently conduct heat to a waterblock if it has themal paste, and a waterblock will more efficiently conduct heat to water if there is a higher flow, AND more turbulence. you wouldnt say that not having a thermal conductor between your CPU and block doesnt matter, would you?

Like i said, your arguing the finer, way more trivial points of this, that really have no effect on the overall output.

And yes, it is simple physics. The point your arguing has to do with changing component order, so your babble about thermal paste or not and what waterblock you use and turbulence, jet inpingment, heat conduction and all that crap, is moot, because for any worthwhile comparison for the effect of loop order, all those factors would be the same across the two test beds. Did i just invalidate your post? The only real (read: the only variables that actually matter) variables are ambient temp, and airflow/fan speed across the heat exchanger. Energy in VS energy out.

I neglected to mention a whole slew of factors, because if your going to have any kind of meaningful comparison, everything would be assumed the same between the two test beds, save for the component order. My bad for assuming you knew how to scientifically perform an experiment.

 

[Erasmus354]

tesseracter: I am sorry, you are patently wrong. Slower flow is *not* better for a radiator and is a common misconception of people who don't fully understand. Instead of trying to actually explain it to you I will just show you the proof.

As you can clearly see the heat dissipated INCREASES as the flow rate is increased.

Now the specific example you mentioned, having two identical radiators placed in parallel as opposed to series, is actually not a bad configuration and I will not argue that. The reason why it is good however is because it reduces the overall restriction of the radiators, thereby increasing flow for the loop as a whole. This improves the efficiency of all the other components in the loop. The drop off for the radiators isn't as bad as the improvement you will get from most of the other components.

However, it is simply not true that slower flow is better for a watercooling flat tube style radiator.

 

[tesseracter]

for any worthwhile comparison for the effect of loop order, all those factors would be the same across the two test beds. Did i just invalidate your post? The only real (read: the only variables that actually matter) variables are ambient temp, and airflow/fan speed across the heat exchanger. Energy in VS energy out.

of course i can deal with the scientific method, and if you would like to duplicate what i did, all you need to do is have a stable ambient temperature(i used an office room), and a stable heat source(I used a TEC). first test the equalized temp with res, pump, rad, block, res; then check with res,pump,block, rad, res. amazingly, the equalized temperature will be different! while you say that everything inside the "closed" system doesnt matter, i contend that they do. Unless you have other constraints regarding the order of the loop(case layout, perhaps) why would you want higher temperatures than you need?

im saying that i can get more energy into the WC system by having a turbulent flow from the pump. you can add that into your energy in, energy out equation.

 

[SolidBladez]

All I can say is that my E6850 never got above 46C on my loop order. And that's with ~1.5V.

Res>Pump>Rad>CPU>Res

 

[tesseracter]

All I can say is that my E6850 never got above 46C on my loop order. And that's with ~1.5V.

and that is better or worse than with the rad/block switched? a single datapoint isnt what we or the OP is looking for.

 

[nissanztt90]

of course i can deal with the scientific method, and if you would like to duplicate what i did, all you need to do is have a stable ambient temperature(i used an office room), and a stable heat source(I used a TEC). first test the equalized temp with res, pump, rad, block, res; then check with res,pump,block, rad, res. amazingly, the equalized temperature will be different! while you say that everything inside the "closed" system doesnt matter, i contend that they do. Unless you have other constraints regarding the order of the loop(case layout, perhaps) why would you want higher temperatures than you need?

You clearly cant if you think because its an office room, its a stabilized constant temperature. May I have detailed notes and a description of the method you used to stabilize the temp of your office? I highly doubt its anything other than an AC controlled by a thermostat in which case the ambient can differ by 10 or so degrees... Unless your office or your pc somehow defies physics, given the same conditions except for loop order, the equilibrium will not be different. I highly doubt the position of said components in the case will effect the equilibrium, although i suppose one may be positioned in an unfavorable place such as a trapped pocket of hot air, but since you have not disclosed your scientific notebook with youre detailed setup and results from your high tech laboratory/office, i couldnt really say.

im saying that i can get more energy into the WC system by having a turbulent flow from the pump. you can add that into your energy in, energy out equation.

Id love to see you actually proove that with an actual exclusion of variables such as ambient temperature in your office...

 

[Erasmus354]

The heat produced by the system is a constant based upon the processors you are cooling. No matter what the configuration of the loop, if they are running at a constant load they are producing the same amount of heat energy. That heat energy has to be dissipated by the watercooling system. The more efficient you cool the processor the lower the temperature of the processor is, but you are still transferring the same amount of heat energy regardless.

The same way that a 120mm radiator will dissipate the same amount of heat energy as a 2x120mm radiator, just at a higher temperature delta. The processor always transfers roughly (accounting for secondary losses through the motherboard traces which can vary slightly) the same amount of heat energy into the watercooling system, more efficient ones just do it at a lower temperature.

Nothing you do will change this fact.

 

[tesseracter]

The same way that a 120mm radiator will dissipate the same amount of heat energy as a 2x120mm radiator, just at a higher temperature delta. The processor always transfers roughly (accounting for secondary losses through the motherboard traces which can vary slightly) the same amount of heat energy into the watercooling system, more efficient ones just do it at a lower temperature.

Nothing you do will change this fact.

Im sorry, are we just getting lost in semantics? my goal is to keep my processor at a lower temperature, and that means the efficiency is of prime importance to me. that means that i want to get heat from the CPU at a lower temp delta. my original argument that turbulent pressured flow from the pump is better than cooler water from the radiator still stands.

the office I used rarely changes more than 2F, it has a big central ventilation system, and during my tests it was rock solid. but even if i did have pictures, charts, and graphs, would you believe me? the cool thing about science is that you can try it out yourself.

 

[Arcygenical]

FYI. I was pumping into my EK reservoir, and then I tried pumping from it directly to the CPU.

Neither made a temperature difference, because flow, in a closed loop, is always constant.

the office I used rarely changes more than 2F, it has a big central ventilation system, and during my tests it was rock solid.

Yeah, I'm inclined to believe this. I've spent days at a time in an office, and the temperature was always rock solid.

 

[nissanztt90]

Im sorry, are we just getting lost in semantics? my goal is to keep my processor at a lower temperature, and that means the efficiency is of prime importance to me. that means that i want to get heat from the CPU at a lower temp delta. my original argument that turbulent pressured flow from the pump is better than cooler water from the radiator still stands.

No it doesnt, at all. Energy in has to equal energy out. Its a law of nature, the law of conservation of energy. Your not getting lower temps because you changed the flow order, period. Theres another variable your overlooking that you didnt keep constant, and its likely ambient temperature.

As I and others have said, the flow rate through out a loop is constant, theres no way its more or less turbulent before or after the pump, especially in a cylindrical tube with smooth walls.

 

[Arcygenical]

Flow is constant within a loop, turbulence and pressure is not. Turbulent flow will decrease (and linear flow will increase) after the momentum of the water stream is broken, and pressure will increase where flow is impeded, and decrease when flow is freed.

Your not getting lower temps because you changed the flow order, period.

Not entirely true in all situations. In a watercooling loop, where the average input of heat is less than 300w, you're absolutely right, the temperature difference will be negligible. But in a loop with a very high heat dump, such as a SLI loop, the very first component will receive a slightly lower temperature, as the temperature of water is slightly lower as well, although not by very much. Cool the cards before the CPU, and the CPU's temperature will be higher.

if you don't believe me, get an MCP355 pump, set up your SLI loop as follows:

CPU > GPU > GPU > Radiator > Pump > Res > CPU

collect temperature data with water following the intended path, then swap the barbs on the pump, to reverse the flow (aka, water through the GPU first, then the CPU, then into the res, rad etc.) and you will notice a small difference...

If your loop holds 250ml, and you have 2 8800GTX's in SLI, which dump 300w into the loop, you'll have an increase in water temperature of roughly 0.3c between the cards and the cpu.

300w = 71cal
71 cal / 250ml = 0.285c delta

 

[nissanztt90]

Obviously when you introduce more components its becomes a bit more complicated, but the heat dissipation capabilities wont change when you change loop order. Like you said, there may be a difference between the GPU and GPUs depending on flow order, but that would be expected since your dissipating more or less heat before a specific block, which in turn can effect temperatures of said block. However, the two GTX's will run cooler since they are being cooled first, and the CPU a bit hotter, and it would be the opposite if the cooler water hit the CPU first. In a multi component loop, yes temps can be effected by loop order (really only cooling block order), but it would be a negligible trade off. I emphasize trade off because, one may be cooler, so the other will inevitably be hotter. Energy in VS out. The extra heat has to go somewhere.

If you were to divide the loop up into the pump, heat exchanger, and water blocks...the order of the water blocks would make no difference at all in the total loop. Once the water gets through all 3 blocks, it will be the same temp wether it goes GPU GPU CPU, CPU GPU GPU, GPU CPU GPU. Before the 3 blocks it will equalize, and after.

Water block order can matter as you said, but theres a trade off. When i said loop order doesnt matter, i wasnt really thinking about multiple blocks, and the other guys arguement was that pumping into a water block vs a radiator or whatever was better somehow, which makes no sense.

Anyway, its really all just trivial pursuit.

 

[Arcygenical]

Well, we obviously agree, but there's nothing like an argument over semantics in the WC forum .