Water cooling- H2O cannot go too fast- Discuss

[mechsiah]

Okay- get your flame-throwers ready. I am about to offend many of you.

I cannot believe that water can move too quickly through an H2O cooling system. While viewing several posts I have seen some people say that the water can move too fast, and that it doesn't have time to pick up the heat.

To this I say "phooey" and then hide under my desk.

My thinking is quite simple (many have pointed this out :) )

No one ever says "Oh, your air is moving too fast in your air cooling system. Turn the fan speed down." Sounds stupid, huh?

While the water picks up less heat, it also takes less time to cool, and is immediately followed by cool water to pick up more heat.

So- Fire away. But go easy...

[StarTraveller]

I must admit that you're brave ;) :D.

This should be an interesting discussion to follow, though. I for one want to hear the arguments for this one :).

[dimmreaper]

Yup, a lot of the die-hards like myself has been preaching this for some time. You really cannot have too much water-flow. However, like aircooling, there is a point of diminishing returns. My endevours have indicated that this point starts at about 200GPH, making anything over 250GPH excessive . . . .

Edit: Actually, I lied, sort of. You can have too much flow, but would have to have like 1000GPH or maybe more to reach that point. With enough flow you will expirience problems with cavitation within the waterblock. Cavitation would reduce the effective surface area within the block, because less internal surface area would be in contact with the coolant.

cavitation- The sudden formation and collapse of low-pressure bubbles in liquids by means of mechanical forces, such as those resulting from rotation of a marine propeller.

[HotDogg]

Well, I'm no engineer but my guess would be that it has something to do with the rate that heat can be absorbed into one medium over the other. Let's say you want to boil water the air above the burner gets hot quicker than the water in the pot. I seem to remember some research on this topic on the procooling.com site.

[mechsiah]

Thanks for the back-up, dimmreaper. And yes on the cavitation, but if you have that problem, you probably need another hobby. :)

HottDogg- As you may have guessed from my vastly technical statement, I too, am no engineer. I have trouble with snap-together models.

That beings said, water is better at conducting heat than air. If it weren't, all of us water cool people would be looking pretty dumb right now.

Your water boiling example may have some flaws- amount of air vs water, temperature of air vs water, etc.

Thanks for your thoughts, though! I'm trying to figure a lot of this out myself.

[HotDogg]

Originally posted by mechsiah
Thanks for the back-up, dimmreaper. And yes on the cavitation, but if you have that problem, you probably need another hobby. :)

HottDogg- As you may have guessed from my vastly technical statement, I too, am no engineer. I have trouble with snap-together models.

That beings said, water is better at conducting heat than air. If it weren't, all of us water cool people would be looking pretty dumb right now.

Your water boiling example may have some flaws- amount of air vs water, temperature of air vs water, etc.

Thanks for your thoughts, though! I'm trying to figure a lot of this out myself.

Yeah, mechsiah I'm no lawyer either but I felt the need for a disclaimer hehe. I'll try to find the research I mentioned then we can all take a gander at some numbers instead of speculating or deduction from flawed analogies.

[HotDogg]

Originally posted by mechsiah
Okay- get your flame-throwers ready. I am about to offend many of you.

I cannot believe that water can move too quickly through an H2O cooling system. While viewing several posts I have seen some people say that the water can move too fast, and that it doesn't have time to pick up the heat.

To this I say "phooey" and then hide under my desk.

My thinking is quite simple (many have pointed this out :) )

No one ever says "Oh, your air is moving too fast in your air cooling system. Turn the fan speed down." Sounds stupid, huh?

While the water picks up less heat, it also takes less time to cool, and is immediately followed by cool water to pick up more heat.

So- Fire away. But go easy...

OK here is a link - seems they conclude that time water is in the radiator matters too. I know there is another site with test data somewhere humm.....

http://www2.apex.net/users/timwhita/flowrates.html

Edit: OK one more analogy - What freezes faster, moving or standing water? This might shed some light on the time in the radiator conclusion. Bottom line is that every system is different so it is unlikely that blanket prescriptions for flow rate can be made. I like the bunker's suggestion at the above link to experiment with a valve to determine optimum flow. Easy to implement and dosen't require calculus.

[dimmreaper]

LOL! The heatercore pictured in that article:

http://www2.apex.net/users/timwhita/cleancore.jpg

Is mine! The pic is taken from an old article of mine on the frontpage, LOL. My stuff has a strange way of turning up in odd places . . .

[dimmreaper]

What they concluded HotDogg was that optimal flow-rate varies with watercooling set-up. Personally, I find the results of that article questionable, as you will notice that with 0.7GPH the radiator water temp is the same as with 0.6GHP, and yet the CPU temp varies by 2°F. That just doesn't sit well with me . . . .

[mechsiah]

Once again, I'm with you reaper. The sample wasn't large and what we really want to see is what happens when the flow goes from .5 to 1.0 to 1.5. Anyways, kudos for the article- it gave us something to chew on.

It did make me think about all the in-hose barbs I've got in my system. I'm probably lucky to be pushing 1gpm myself. HA!

[HotDogg]

Originally posted by dimmreaper
What they concluded HotDogg was that optimal flow-rate varies with watercooling set-up. Personally, I find the results of that article questionable, as you will notice that with 0.7GPH the radiator water temp is the same as with 0.6GHP, and yet the CPU temp varies by 2°F. That just doesn't sit well with me . . . .

Yeah dimm I can't disagree with you the setup is questionable (not to mention that they couldn't take their own picture apparently).

Sorry I can't for the life of me find the other link but the findings were similar, if memory serves, increased flow improved cooling to a point where further increases degraded cooling.

The graphic they show does indicate this trend, however, as you correctly observed, the small flow changes don't properly show the full flow vs. temp function.

I have no further analogies so I'm done for now.... unless I find that other elusive link;)

[Baboo]

(I've a new log-on name. It's been a while since I've been here. First time I've been to the 'new' vboard!)

Here's my pennies.

I'm currently cooling with a Swiftech water block, Senfu Twin fan radiator (with 2 80mm deltas) and an Ehiem 1046 pump. I borrowed a Ehiem 1048 and put it in my system. I saw no difference in temps.

Monitoring chip temp with a DD5 and Senfu probe (both read the same temp). Also monitoring radiator inlet and outlet temps (as well as other things).

[BillA]

lets deal with numbers:

http://www.ecom-answers.com/OCeg.wmf

a couple of notes:
this is a poor 3d graph using some "bad" data generated while I was/am learing how to use the program
(the dips and peaks indicate "off" calibration of the flow sensor)

(good plots with good data will be in the "Radiator Heat Dissipation" article in progress for JoeC)

the scale of the X axis is compressed and actually represents data collected at flow rates from 0.24 to 2.08 gpm
due to this compression the slope is not too apparent, but the heat dissipation does (slowly) rise as the flow is increased

this graph says MANY things, but regarding the coolant flow it is clear:
the "laws of physics" (as we use them) have not changed
greater flow increases the radiator's heat dissipation

now how much, and at what relative cost are very different questions

incidentally, the measurement of all this requires equipment quite beyond anything now in use by "reviewers"

be cool

[dimmreaper]

Thanx for sharing the kick-a graph BillA . . . .

[r0ckstarbob]

Originally posted by HotDogg

Edit: OK one more analogy - What freezes faster, moving or standing water? This might shed some light on the time in the radiator conclusion.

standing water freezes much faster then moving water. in fact, if the more the water moves, the farther the freezing point drops. yes it is possible to get subzero tap water if you move it quickly enough.

dunno if this was a rhetorical question or not, but figured i'd field it anyhow.

this is interesting stuff. i'm definately curious myself.

[HotDogg]

Originally posted by r0ckstarbob


standing water freezes much faster then moving water. in fact, if the more the water moves, the farther the freezing point drops. yes it is possible to get subzero tap water if you move it quickly enough.

dunno if this was a rhetorical question or not, but figured i'd field it anyhow.

this is interesting stuff. i'm definately curious myself.


Yes, this matches my experience; moving water is slower to freeze in comparison to standing water. However, I'm not at all suggesting that zero flow would be optimum. I can't question the graphic the other poster provided since I find it difficult to think in 3 dimensions. It is quite possible that for the range of flow rates that a watercooled setup might have that increased flow will increase cooling ability. Clearly more research is needed. I'd be happy to do it but I just don't have the equipment.

[r0ckstarbob]

um, just a clarification. i think theres some confusion here.

the example of standing water or moving water isn't a good example for this thread, though a good attempt.

moving water does not warm it up or prevent it from getting cold. it merely prevents it from freezing like standing water does.

you can actually get COLDER water then frozen water if it's moving.

standing water will freeze at 32F. moving water (depending on how much it moves) can reach temperatures well below that.

[TT120]

While it may not be possible to move water too fast through the block to absorb the heat, it is possible to move it faster than the radiator can dissipate it. The water has to have enough time in the radiator to lose all the heat it has picked up in the block.

I learned this first hand with my Camaro. It was running a little hot and I figured that I would remove the thermostat to allow for better flow. Well, it flowed better alright, the water didn't stay in the radiator long enough to cool off and my car ended up running hotter. I used a hose clamp on the hose to restrict the flow and the temps came way down.

Now, I understand that a car is not a computer but the cooling system is the same, albeit much smaller, the principal is the same.

[r0ckstarbob]

that is an excellent explination. so while it's possible to improve heat transfer with increased flow up to a point, you'll eventually have to either increase your reservoir or your radiator to continue to see improvements right?

so the bigger the water reservoir, the faster you can pump your water - the faster you pump your water, the more performance you can see from your watercooling setup.

sound about right?

[TOMATOMAN]

the way i see it if the water moves fast it stays less time in the block so it picks up less heat so it has to stay less time in the radiator.

[-=HN=- Wild9]

so wait, you would have to get a bigger resivoir, pump water faster, and get a bigger radiator, to hold the water for a longer time to dissapate heat, to get better performance?

[IFMU]

Ok now most of this is making sense... and I can see the points made on all the sides involved...
Now, you need the water to remain within the rad longer to remove the heat that was collected in the block... but you want it to run at a decent speed within the block to collect the heat... now basiclly you want a fast pump to push it through... now what I was thinking... Would you be able to get a fast pump to push the water first into the block... (again at a high rate) then when it leaves the block... would there be a way to slow down the speed so it would need to remain within the rad longer? then increase the speed again afterwards maybe? Now understand that within a closed system I dont think that would be a decent possiblity... However... what if you were to place a res just before the rad so the water would have to reach it.. come to a near stop there.. then be slowly pushed out of it into the rad? so the speed there would be much slower than going through the rad... the on the opposite side of the rad (outgoing) you place the pump.... so the speed would be replaced again (the water would be forced to move at a higher rate again?)

Understand what I mean? I see it but not sure if Im explaining it right...

[BillA]

disregard the reservior, it's like a thermal shock absorber; contributes virtually no additional cooling

the velocity is easy to manage
even if you don't know your flow rate (and if you don't, measure it)
the parameter of interest is the relative rate of the wb and rad
this is in proportion to their volumes (2x the rad volume as the wb = 1/2 the velocity)

measuring by weight is the easiest

all other things being equal - which they never are
a higher volume rad is always to be preferred

be cool

[r0ckstarbob]

nooooooo silly. don't just disregard the reservoir.

bigger reservoir or bigger radiator. one way or the other you'll have to go bigger. which one is preferable? i dunno. is probably a discussion for another thread but basically it's going to depend on your application and your setup. running the water through the radiator first and dumping it into the reservoir wating to get pumped will work just fine using the "thermal shock" method. the water in the reservoir is still cooling down. in fact, until that water hits the CPU its still cooling down. and the minute it leaves it, it's cooling down again. the more time away from the CPU that water has, the more time it has to cool off, radiator or not.

you're right - a larger, higher quality radiator will probably much much more efficiently but will cost more money. it will also put more strain on your pumps and reduce pumping speed in which case you're sunk until you get a harder working pump to take advantage of your new and improved radiator. and your new superpump is going to cost even more moola.

it's a balance issue of what you're willing to sacrifice. i doubt we'll find the perfect one size fits all technique, but at least i think we've answered the question about pump-speed vs heat transfer efficiency and perhaps forwarded at least a few theorys of how to put it into practice, no?

[r0ckstarbob]

Originally posted by ifmu
would there be a way to slow down the speed so it would need to remain within the rad longer? then increase the speed again afterwards maybe?

you bet, it's definately possible. as long as you understand that the fastest that water is going to travel is going to be restricted to the speed it can reach in the narrowest part of the system.

follow me.

you're using 3/8 inch hosing and 3/8 inch waterblocks. if the hoses in your radiator were bigger then your tubing (say 1/2 inch or even 3/4 inch) the water would slow down when it hit the radiator, stay in the radiator longer, and then speed back up once it hit the outlet nozzle on the other side which would be at 3/8 inch again. it would definately work.

[Flash]

OK, you guys have lost your minds. The water is going to flow through the block and the radiator at the same speed, period. Using a reservoir makes no difference in terms of speed. To lower temperatures you need to increase the surface area or airflow throught the heat-ridding component, i.e., the radiator, not the reservoir.

Having said all that, I can't figure out how to determine the ideal flow, or even if there is such a thing. In my experience (same system using 80gph, 350gph, and 700gph pump) the amount of flow doesn't make much difference.

[mechsiah]

OK, I wasn't expecting this thread to resurface, but here it is.

First off, BillA (and others) are correct about the resevoir. It does not contribute significantly to cooling, only the initial amount of time it takes the water to heat up (and cool down when the machine is off). Granted, it acts as a very bad radiator, but this is negligible.

Secondly, TOMATOMAN nailed it. If your water is moving slow, it will pick up more heat in the block and NEED more time to cool down. The reverse is true as well.

A good radiator system is the key. If the radiator is good, then faster is better- with a few extreme exceptions.

[BillA]

for those with a very serious interest (compulsion ?) in flow, try this thread (18 pgs)

flow rate, faster, slower, not important (http://hardforum.com/showthread.php?threadid=142512&perpage=15&pagenumber=1)

Flash,
no, I know where MY mind is
perhaps you are confusing flow rate, velocity, volume, area, and ?

for a given pump's output (gpm);
if one doubles the cross-sectional area of the pipe
one WILL half the velocity (fpm)
therefor
if one has twice the volumetric capacity in the rad as in the wb
one WILL have half the velocity

the dominating attribute for rad performance (on the liquid side)
is the ratio of internal volume to wetted area, and contact time

which is why
all other things being equal - which they never are
a higher volume rad is always to be preferred
and flat tubes (with soldered or braized fins) are always more efficient than round ones

be cool

[nihili]

Originally posted by r0ckstarbob


follow me.

you're using 3/8 inch hosing and 3/8 inch waterblocks. if the hoses in your radiator were bigger then your tubing (say 1/2 inch or even 3/4 inch) the water would slow down when it hit the radiator, stay in the radiator longer, and then speed back up once it hit the outlet nozzle on the other side which would be at 3/8 inch again. it would definately work.

Of course in doing this you also decrease the surface area of water exposed to heat transfer, so the radiator becomes less efficient.

nihili

[nihili]

Ok, this is a longish simplified explanation of how flow rate should effect water cooling systems. Let me know if I've got something wrong.

nihili

Ok, let's start with an open system. For simplicity I'll be ignoring the heating effects of friction, viscosity, etc.. You’ve already chosen your water block, you have an infinitely powerful variable flow pump, and two hoses. One hose leads from Lake Michigan, through the pump, to your waterblock. The other leads from your waterblock to the Mississippi river.

Q: How fast should you set your pump?
A: As high as you can.
Reason: The rate of heat transfer between the block and the water depends almost exclusively (once other parameters such as composition and structure of the block are set) on the difference in temperature between the water and the block. The slower the water moves through the block, the higher it’s average temperature becomes, and hence the slower heat is transferred from the block to the water. Even if you have a flow of 5 gazillion gpm, you’ll get better cooling if you turn it up to 5 gazillion and one. Of course the returns diminish. The difference between 1gpm and 2gpm is huge, the difference between 1,000,000 gpm and 1,000,001 gpm is small. But there is always a difference, however miniscule.

Ok, so everyone starts noticing that the water level in lake Michigan is dropping at an astonishing rate, and one day these men in long black coats show up at your door and gently suggest that you connect the ends of your two hoses together since it would be a crying shame if you were to get sucked into the intake of your own pump. Cooperative sort that you are, you take their suggestion and connect the free ends of the two hoses.

You now have a closed system. You turn on the pump and the temperatures start to climb. The problem is that you keep putting heat into the water, but you don’t take any out (yes, these are super-insulated hoses, now go away.) So the water will continue to get hotter and hotter until it equals the maximum temperature of your cpu. If you have an Athlon, that means about 372 celsius if I remember the video correctly. Not wanting to fry your cpu you hit upon the idea of putting a radiator in between the two hoses.

Now a radiator works on the same basic principles as a water block. Once you set the parameters such as structure and composition (i.e. once you choose a radiator) the rate at which the radiator can dissipate heat is determined by the difference between the internal and external termperatures. The higher the difference, the greater the rate of heat transfer. The reasons you have fans blowing on the radiator is to keep the air around it from heating up to much and decreasing the rate of heat transfer. Now, if your sole goal is to maximize the rate of heat transfer, and you had an infinite supply of hot water, then you would just turn up the pump all the way again. But it’s not that simple now.

When you increase the flow across the water block, you increase the efficiency of the heat tranfer, but in doing so you decrease the average temperature of the water exiting the block. So, higher flow means, lower average exit temperatures from the block. But that in turn means a lower internal temperature in the radiator. So the radiator becomes less efficient. Basically, as you increase flow, the water block becomes more efficient, but the radiator becomes less efficient. As the radiator becomes less efficient, the intake temperature of the water block goes up, so the cpu gets hotter.

The problem is that you’re trying to use a single variable, flowrate, to control two other variables, radiator and water block efficiency, which are inversely related. You want to maximize the efficiency of the system, but when you make one part more efficient, the other part becomes less efficient. As I’ve already alluded to, efficiency curves aren’t straight lines. And the efficiency curves of one system will be different than the efficiency curves of another system because of differences in composition and structure of the heat transfer points (I’m still ignoring friction, resistance, etc..) So there is no one correct answer to how many gpm you should use. Sometimes increasing the flowrate will decrease the overall efficiency of the system because it will decrease the radiator’s efficiency more than it increases the water block’s. Since the efficiency of the radiator is also influenced by the temperature of the air blowing over it, different flowrates will be optimal for a given system on different days. It’s even possible that for a given set of conditions there are several equally optimal flowrates.

So, what should you do. Get a system that can handle as much flow as you can afford. Put in a valve so you can adjust and tweak the flow rates to optimize your particular system in it’s particular conditions. Then start looking for how to change other variables to increase the efficiency of the overall system.

[r0ckstarbob]

Originally posted by nihili


Of course in doing this you also decrease the surface area of water exposed to heat transfer, so the radiator becomes less efficient.

nihili

guess that depends on the shape and design of your radiator piping. but yup, yer right. forgot about that. was just thinkin about flow rates, forgot about surface area on a standard round pipe. still you're reducing the load on your pumps, and bigger piping means better water turbulence inside the piping which equals better heat transfer... theres a load of other things so it is possible to balance that out. it is possible though, depending on the design, to have less surface area inside the piping but have better cooling from your radiator. it all depends on the design. you could shove a coiled copper wire down the length of your radiator tubing... use wider flatter piping in the radiator... etc etc etc...

[r0ckstarbob]

okay, something i should explain

i'm planning on using a radiatorless system and am actively cooling my 3.6 gallon methanol/water copper reservoir to -30F. this is a closed system, not a bong system.

please feel free to substitute the word reservoir with radiator at any point in any of my posts. for me, all i worry about is a reservoir. for all y'all, you worry about radiators. as i said, different setups, different applications, different concerns. believe me, i'm well aware of the performance differences of a radiator versus a reservoir. at subZ's though, when the air is almost always going to be warmer then the case (and the evaporator coils), actively cooling the reservoir makes alot more sense then trying to incorporate a radiator, believe me, so thats where my focus is.

[Burning Phoenix]

Why doesn't someone here try using a 4 core car radiator and house fan and see if it works any better. I would like to see pictures of that one and how would you explain to it to the people you live with? Think they would think you were crazy. Would be funny though. A desk with a radiator attached. LOL