Cooling Technologies Explained Discussion

[Joe]

With an article of that size and depth there are bound to be some slip ups

"quote:
--------------------------------------------------------------------------------
The Coolant in all cases has picked up ~90% of the heat from the heat source.
--------------------------------------------------------------------------------
How do you arrive at the figure for this and parasitic heat loss? Are they guesstimates or calculation-backed? "

Half guestimation and half educated guess I discussed this topic at length with a few geeks from different places. One is a thermals geek who works at GM working on Engine block design, mainly coolant paths. He had some interesting numbers about car engines and how much heat they loose to the surrounding air compared to what goes through the radiator. For a car engine its around 15% by their calculations at a certain temp Delta from the ambient ( don’t know what that delta number was). After we talked and discussed the thermal densities and the environment that the cooling system runs in we came to a round about number of 5 - 10%. The fact is the parasitic loss is there, the EXACT number is almost impossible to figure out with testing gear most geeks have access to. Also it will change by how cool or hot the ambient air is, and air flow, and a billion other factors.

"quote:
--------------------------------------------------------------------------------
In water cooling I would venture a guess that a good deal of heat is lost in the hoses, reservoir, and any places its in contact with a case side or something.
--------------------------------------------------------------------------------
Sorry, nitpicking again, but didn't you previously establish this as being 10% of heat from source? "

Yep! the extra 2 lines of text didn’t take to long to download I hope.....

"quote:
--------------------------------------------------------------------------------
So with that said the Radiator on a H2O rig moves a lot of the heat but has a much easier job of it since there is a much higher mass between it and the core to loose heat at different stages.
--------------------------------------------------------------------------------
If so much heat is lost along the way, why do you need a radiator in the first place? If 10% is lost along the way, there is still 90% for the radiator to dissipate. It may be somewhat easier, but I think much is stretching it a bit far. "

Well compared to how many places a HSF has to loose heat... You need to look at the bigger picture in the writing not one sentence at a time.

"quote:
--------------------------------------------------------------------------------
The water coming back from a radiator on a H2O rig really may not be much cooler than the incoming coolant into the radiator. This is a factor of how effective the radiator is for the volume/speed/type of coolant you are moving through it. Of course the closer to ambient the better.
--------------------------------------------------------------------------------
I think it's important to note at this point that the difference in temperature across the radiator itself is going to be actually rather small. Water in the entire loop should ideally be at as similar a temperature at possible, as that is the temperature at which heat loss in the radiator(and along the way) exactly matches heat gain in the waterblock."

Isn’t that what I said... Just not as over worded?

"quote:
--------------------------------------------------------------------------------
Pump: Pumps don’t mind warm coolant, and introduce a few watts of heat themselves to the coolant, so its good to have it BEFORE the heat exchanger.
--------------------------------------------------------------------------------
This is a point I always like to contend. Its not that water will suddenly jump a few degrees between before and after the pump. It's that the pump adds another heat source to the water, raising the temperature the water has to be at to dissipate the heat. I lay before you a challenge, measure the water temperature before and after your pump. Is it even perceivably different? I'll venture as far as saying no.
What does this mean? It doesn't really matter where the pump is, the deltaT across it is negligible. It's not a sudden rise in temperature across it, it's the added heat that it contributes that raises the equilibrium temperature. This implies that the pump can pretty much go anywhere. "

Maybe you missed a line that I wrote:
"Just as with any other cooling, the movement of the coolant is vital to how well the entire system works. This is a topic that many people differ on, and some claim it makes no difference in reality for our cooling systems. That may be true, but if you are designing a cooling system, why not do it right?"

the difference may be VERY SMALL... But why would you spend that much money and time on a cooling system just to design it half assed?

Yes there are some nice grammar issues in this one I got some fixes to do tonight on it hehehe

It wasnt meant to teach old dogs many new tricks, it was meant to act as a base level for people to see what technology does what and where they may be able to use it. glad overall you liked it!

[pHaestus]

Joe your comments on designing loops and Marco's disputes about the importance of positioning the components I think might only come in conflict in people dealing with small cases. As the trend moves to water cooling in smaller and smaller cases, there are a lot of tradeoffs to be made to fit everything in. No it isn't perfect to put a pump right before a block instead of right before a radiator, but it might allow you to get the overall loop a lot cleaner than putting it together a more technically sound way.

I was also glad to see your distaste at radiators on the top of cases; I don't even consider "top of the loop" when putting water cooling in a case. I just have never had luck keeping them airfree that way.

[Joe]

heheheh my distaste

Its just logic, personally I think teh top of a case makes a great location for a radaitor if you can some how keep the reservoir and air trap over it.

One logical reason not to put a radiator in the top of a case is, Heat goes up. Why you want the hot air in your case to go up and through your radiator?

and Yes I agree on tradeoffs like that have to be made some times... I just cant ever seing myself settle for that though in one of my systems... I am just weird that way.

[meck]

oops brian256 i thought a capliary tube is to be inserted into the evaporator core to allow a expansion valve to throttle .in fridges i thought they used fixed orfice tubes. tev could you break that down for me please thank you

[superart]

I got a question about the phase change topic. In the article it was stated numerouse times that before building your own setup, it would be prudent to get HVAC certified. I have a question about that. Where would one go to get this certification? Are there age restrictions? How much is it? and this may be really dumb, but how is it pronounced?....
(H-V-A-C) or H-vac (as in vacume cleaner)

I have read a couple articles about phase change setups, and all of them say that getting HVAC is a good idea, but no one gives any details about the course/certification itself. I would really apreceate any people that are already HVAC certs to shed some light on this.

Thanks

[Brians256]

meck,

Capillary tubes are a fixed orifice junction between the condensor and the evaporator, but they are not a valve. Valves can be adjusted; capillary tubes cannot be adjusted. Consumer refrigeration systems (for homes) use capillary tubes for expense. Commercial systems (for restaurants and other big sites) might use a TEV (thermal expansion valve). In a large setup, the efficiency of a TEV might pay for the extra initial expense in a short time.

superart,

I have always heard HVAC pronounced like H-Vac ('H' as in the letter then Vac as in vacuum).

I really stress the HVAC certification because of the legal liability. I cannot promote what could be an illegal action. The legal landscape today makes me err towards caution. However there is another reason. In theory, you could probable build a system from the description that I gave in the article. But, an HVAC certification gives you practical experience in charging a system correctly. For instance, you can charge a system by watching the amperage draw on the compressor, or by watching the superheat temperature, or by a number of different methods. I also think that there is paperwork to fill out for certain operations like using R-12.

I am not HVAC certified, unfortunately. Last I checked, the easiest way to find out is from your local repair shop. The refrigerator manufacturers are uninterested in dealing with individuals, for the most part. Alternatively, you might contact the Air Conditioning and Refrigeration Institute (ARI). Most HVAC certs are issued as part of 18-month or longer trade schools, but there might be alternatives in your area. You just need to check.

[Joe]

I just updated the entire article with a list of changes Brian pointed out! just spelling and such.

[Brians256]

Wow! That was fast, Joe! One important thing to note... in spite of my nitpicking, I was unable to find any technical flaws. For the sheer amount of territory covered, Joe, pHaestus and Brad all produced remarkably error free overviews. Kudos! They really do know what they are talking about.

[Marco]

Originally posted by Brians256
Do you mean that I did cover compressors enough or not? It looks as though you started to comment before reading the article all the way through. I hope that I did cover it well enough; let me know if you do have some questions. As for the sponge metaphor, it just seems natural, doesn't it? Glad you liked it!

Also, your proofreading needs some work. You missed at least two grammar errors in my portion, and (I think) two spelling errors in Joe's portion.

Yeah that's exactly what I did, I was pointing out stuff as I read. I mention that at the top of my post

Your coverage is more than adequate, its the coverage on the "intro/primer" page that is confused.

As for later grammar/spelling errors, I'll confess that as I realized how long the article was a switched from reading to skimming. Time constraints yknow

[Marco]

Originally posted by Joe
You need to look at the bigger picture in the writing not one sentence at a time.

Hey, I'm just trying to help by making things as clear as possible.

Originally posted by Joe
"quote:
--------------------------------------------------------------------------------
The water coming back from a radiator on a H2O rig really may not be much cooler than the incoming coolant into the radiator. This is a factor of how effective the radiator is for the volume/speed/type of coolant you are moving through it. Of course the closer to ambient the better.
--------------------------------------------------------------------------------
I think it's important to note at this point that the difference in temperature across the radiator itself is going to be actually rather small. Water in the entire loop should ideally be at as similar a temperature at possible, as that is the temperature at which heat loss in the radiator(and along the way) exactly matches heat gain in the waterblock."

Isn’t that what I said... Just not as over worded?

I guess I failed to express what I meant clearly. Is closer to ambient better? If it's closer to ambient, then conversely that means that the flow rate is lower and the water exiting the block is at higher temp. This means that the water goes thru a greater temp rise in the block. Averaging the temp in the block, it will be the same as if the flow rate had been faster, and thus the temperatures on the "hot" and "cold" side of the loop been closer. So how is it better when rad output is closer to ambient?

Originally posted by Joe
Maybe you missed a line that I wrote:
"Just as with any other cooling, the movement of the coolant is vital to how well the entire system works. This is a topic that many people differ on, and some claim it makes no difference in reality for our cooling systems. That may be true, but if you are designing a cooling system, why not do it right?"

the difference may be VERY SMALL... But why would you spend that much money and time on a cooling system just to design it half assed?

If it makes no difference then however you do it, it is still right...
There is nothing half-assed about putting the pump anywhere in the loop. What would be half-assed would be getting a weak pump.

That said, I agree with what pH has to say about small cases. If it doesn't increase the difficulty of putting the system together, go ahead and put the pump wherever you want. However, if trying to put the components in a specific order gives you some technical problems, keep in mind that the effect it has is virtually negligible, so there is no need to lose any sleep over it.




You've addressed all the more subjective issues I raised, but how about the more technical ones?

In any case yes I did like the article, what I'm trying to do is just try to help to improve it by voicing my views for discussion. Please don't take this as me trying to attack your work, I am not!

[Joe]

Hey Marco,

Don't get me wrong I am not bent about your questions at all! I like them actually, as they are adding some parity to the article.


As for the whole "pump placement" deal. To me it comes down to this:

Your pump DOES add heat. it may be a very small amount, but you cant deny that the pump transfers some heat to the coolant. Now in a optimum situation, where would you want the coolant to go next? right to the thing that you are trying to cool? or to a heat exchanger to remove that heat before it goes to the thing you want to cool?

Logic would dictate that you would want the lowest temp coolant to go to the object you are cooling, so you should have a heat exchanger right before that object to remove any heat that the system as absorbed in its loop before that.

Also some pumps DO generate a good deal of heat. The Hydrothruster is a good example, as well as a few of the other high end pumps. I did a test a few months ago on my Hydrothruster and at full bore with no fan on the radiator running in a loop with no CPU the water heated up to around 100F in 30 min and would have gone higher, with a small fan on the radiator it dropped 10 DegF to 90F with an ambient temp if 73F. With a 120mm fan the coolant remained at around ambient or a degree or so above.

This shows that the pump CAN put a good amount of heat in the coolant, if you have something that is pre heating the coolant before the core wouldn’t you want to have that coolant cooled off more before it hit the core?

You can do what you want and in some cases its easier to not do it that way... I just see it as a main aspect to consider when designing a system. To some people the "reasonable" amount of heat that the pump may introduce to a system is not an issue, and too much work to worry about. But to me the systems overall design is by far one of the issues I look at the most when planning it around a case.

As for the radiator ambient temp deal. Isn't the output of the radiator always better if its at ambient or as cool as possible? Yes i know keeping a higher flow rate through the block will give an overall more "averaged" heat up cycle to the coolant, there will be a very small temp increase through the WB, and a very small temp drop through the radiator at a higher flow. But in reality having a radiator output ambient temp coolant to the core is a best case scenario. It all depends on how you tune your system for the right flow rate for the components you are using.

[Dix Dogfight]

I loved the article. It covers the most of the basics and is really well written.
But what is this talk about "heat soak point".
There is NO such thing. The increase in heat is linear when a constant amount of energy is added. Untill it changes phase of course. But i guess the CPU will fry before the copper/alu melts.

[benrich]

the part about watercooling was very helpful. but i'm a little worried. i just ordered the innovatek water cooling kit recommended by http://www.overclockers.com/articles545/ and its my first journey into watercooling. you say i need the res to be the highest point, but on the kit the res connects directly to the pump. do i need figure out a way to mount the whole assembly in the top of the case? i understand i could add another res/bleed point, but i don't really consider that an option since i already payed $225 for this kit and it comes with a real nice res as it is.

also, about the water, antifreeze, and water wetter. you say theres no reason to put antifreeze in, and you recommend water wetter. but in one of your older articles you recommended purple ice. and in the review of the kit at overclockers.com he says "Note that any design using dissimilar metals requires that the cooling fluid contain anti-corrosives; anti-freeze is a good additive - a solution of water with perhaps 10-15% antifreeze is fine."

should i use water wetter or purple ice, and should i combine it with a little antifreeze or not?

tia

[Joe]

The reservoir can be at the bottom of the case, its an air trap that should be at the top.

a Pump/rese combo should always be at the bottom of a case (the pump should be the lowest point in the system, something I forgot to talk about I guess)

Hope that helps, your setup will work just fine. and I would use water wetter and not antifreeze.

[Haddy]

Originally posted by Joe
a Pump/rese combo should always be at the bottom of a case (the pump should be the lowest point in the system, something I forgot to talk about I guess)

y is this...

[benrich]

Originally posted by Joe
The reservoir can be at the bottom of the case, its an air trap that should be at the top.

thanks for the quick response. but uh... i'm sure your familiar with the kit, the rese is the air trap for this system. are u suggesting i add yet another component to this already overpriced system?

sorry if i'm missing something obvious here :shrug:

[Brians256]

Airtraps do not need to be expensive. Want to see mine? Yes, I'm upgrading, but this one was built, and time is limited, after all.



That one cost about $2.05, including the expense of the brass barbed fittings.

[Joe]

my example of an air trap was a "best case" type deal. You CAN use other stuff, like a Pump/Rese on the bottom of the case and get similar results just not as instant. ( may take a day or so for all the air to get out of the system and such).

Joe

[Marco]

Originally posted by Joe
a Pump/rese combo should always be at the bottom of a case (the pump should be the lowest point in the system, something I forgot to talk about I guess)

If it's a submersible pump in the res it doesn't need to be at the bottom. As long as there stays water in the res keeping the pump submerged it should pump fine. Issues arise if you have it right at the top, and air pools and gets sucked into the pump.

It's with inline pumps that trying to keep the pump at the bottom helps, because air has the bad tendency of ending up in the impeller, killing its efficiency. I saw a system here with an eheim pump put in with the intake underneath and the output at a side The pump was about at midheight in the loop, and I was asking myself, how the hell did that guy keep the air(edit:air not water!) out?

In any case here is a solution I was pondering for inline users. This is particularly relevant to me with a bong, that can suck air on a weekly basis. Manual priming of the pump can be complicated, as air can pool in countless places in long bong lines. So I was thinking of putting a small cheap submersible pump in the res, and turning it on only when the system needs priming. Guess this does not really apply to closed-loop resless systems though.

[Joe]

Yep in some ways you are right, it comes down to Centrifugal pumps CANT suck water, they can only push water. Using that logic you can do what ever you want as long as your pump isn't trying to pull water up from a pool below.

[Marco]

Originally posted by Joe
Your pump DOES add heat. it may be a very small amount, but you cant deny that the pump transfers some heat to the coolant. Now in a optimum situation, where would you want the coolant to go next? right to the thing that you are trying to cool? or to a heat exchanger to remove that heat before it goes to the thing you want to cool?

Logic would dictate that you would want the lowest temp coolant to go to the object you are cooling, so you should have a heat exchanger right before that object to remove any heat that the system as absorbed in its loop before that.

Also some pumps DO generate a good deal of heat. The Hydrothruster is a good example, as well as a few of the other high end pumps. I did a test a few months ago on my Hydrothruster and at full bore with no fan on the radiator running in a loop with no CPU the water heated up to around 100F in 30 min and would have gone higher, with a small fan on the radiator it dropped 10 DegF to 90F with an ambient temp if 73F. With a 120mm fan the coolant remained at around ambient or a degree or so above.

This shows that the pump CAN put a good amount of heat in the coolant, if you have something that is pre heating the coolant before the core wouldn’t you want to have that coolant cooled off more before it hit the core?

I'm not disputing that pumps can produce a good deal of heat. The thing is that the heat from the pump does not only heat up the water between it and the heat exchanger!

In a system, heat adding components are the waterblocks and the pumps. Heat loss components are radiators and the lines themselves. Now, the heat loss components are passive, meaning that they don't move heat by doing work like a peltier does, they remove heat by optimizing conditions for convection.

What does all this mean? This means the system must reach a temperature where the heat is lost at the exact same rate as it is gained.

Why is this? This is because the rate of heat gain is constant, but the rate of heat loss is proportionate to the temperature difference between the heat transfer medium and the ambient. That means the hotter the water the faster it will lose heat, given ambient temperatures stay constant.

Now, returning to our problem, what happens is that the pump's added heat requires water in the entire loop to heat up, in order for the rate of heat loss to be able to match the rate of heat gain.

So in summary, if the pump raises the temperature of all the water in the system by an equal amount, what difference does where the pump itself is make?

As for the radiator ambient temp deal. Isn't the output of the radiator always better if its at ambient or as cool as possible? Yes i know keeping a higher flow rate through the block will give an overall more "averaged" heat up cycle to the coolant, there will be a very small temp increase through the WB, and a very small temp drop through the radiator at a higher flow. But in reality having a radiator output ambient temp coolant to the core is a best case scenario. It all depends on how you tune your system for the right flow rate for the components you are using.

For a radiator to output air at or near ambient temp, either the quantitiy of heat that it is removing is extremely small, or the flow rate through it is extremely slow. I don't think we even need to address the former case . In the latter case, the water may enter the waterblock at the lowest temperature. But what temperature does it leave at? The slower the water flows, the more heat each unit of water will have to absorb in its flow through the waterblock. So it may enter somewhat cooler, but if it exits substantially hotter, it means that it will be hotter in some parts of the block, and in these parts the heat removal rate will be lower. On the average, in theory, the results should be about the same.

However, all this does not take into account some factors. Maybe we could discuss them a bit. One factor is the water in over core issue. One could presume that this would work well with lower flow rates where water comes in cooler, but we also know that it will be heated more intensely for longer. What happens as it heats up?

Another factor is the change in properties of the thermal barrier due to increased turbulence at higher flows. So is any gain acheived by "cooler" water entering over the core greater than the loss incurred by the turbulence due to loss in speed?

[EMC2]

* screams and runs from the thread *

[Volenti]

Yep in some ways you are right, it comes down to Centrifugal pumps CANT suck water, they can only push water. Using that logic you can do what ever you want as long as your pump isn't trying to pull water up from a pool below.

No, thats' wrong, thay can suck water just fine, they just can't suck air very well, and won't self prime.

*rushes out to shed to perform an expriment*

My spare 2500L/H Centrifugal pump (once primed) can easily suck water up 2 metres of 1/2'' tubing and spit out a decent stream.

Of course you wouldn't want to mount a pump like that in a mission critical suitation, since it only has to hickup and the water column will fall and the pump will be left trying to pump air in vain.

[Brad]

I've done that with my pumps. of course they won't suck air up. I filled my sink up with water, turn the pump on in it, with a 2m long hose to the inlet of it. Just lifted the pump out of the water, and it did suck the water up, and pump it out the outlet.

[Joe]

Originally posted by Volenti


No, thats' wrong, thay can suck water just fine, they just can't suck air very well, and won't self prime.

Believe what you want, but you will cut your flow WAY down, and ANY... ANY little disrpution in the coolant flow and the entire thing stops pumping water where it should. A sealed closed loop system is a bit different, but an open rese below the pump is pretty much about worst case scenario.

I prefer to NOT run something that is just relying on water tension to do. ( I mean in some way its just a fancy siphon then.) Which if you think about it.. water wetter could make that even harder to do when it breaks up the surface tension of the water.

Read the the packaging for the pump in there it says pretty clearly, that the pumps are NOT designed to suck but only to push coolant.

Also the higher end the pump teh easier time it would have "sucking water" as it would be able to overcome the gravitational forces that are pulling the water back down. a lower end pump may not dig it at all if there is much resistance on the output side of the pump. Just cause it can be done doesnt mean its the right way to run something.

For all intents and purposes they CANT suck water on their own. (needs to have a water stream established Then raised above the water surface ( prime it)).