Is too much water flow a bad thing?

Because I am planning to build an external water setup similar to those of Naja02, i'm curious to find if a more powerful pump could damage water blocks? IE if I buy a garden water pump to push the water around, am I in danger of ruining the waterblocks because the pressure is higher or because the flow is faster? Do these things matter or is flow just flow?

Flow itself is generally no issue.

Pressure, however, could cause things to stretch or crack.

Originally Posted by Down8

Flow itself is generally no issue.

Pressure, however, could cause things to stretch or crack.

The Iwaki RD-30 pump some people use for WC shows signs of inflating the tubing a little just because of the sheer pressure.

so to rephrase my question, then is too much pressure a bad thing? if i spread the pressure throughout say RES>PUMP>3xRAD>Y>CPU+CPU>CHIPSET+CHIPSET etc

what i was thinking was submerging three rads in antifreeze then going to the cpus, chipsets, gpus, etc and if the pump was strong enough i could place all of this say next to my washing machine when is on the other side of my basement then just pvc pipe the lines over to the side where the computers are and then obviously have the return line do the same, so say no more than 50 feet combined both directions (15 feet to the computers allow 5 to 10 foot for internal tubing and back to the return line then 15 foot return line)...would a 500 or more gph pump from lowes be too much pressure on the system is my question.

500gph! Wow.

So, you want to have a liquid to liquid heat exchange. That may be smarter than liquid to air.

One loop is the basic liquid computer cooling loop while the other is the PVC and tank holding the rads coming from the computer.

Sounds feasible. Too much flow would hinder the heat transfer from the rads to the other liquid loop.

Maybe 250gph would be a better rate.

Once you go too high in flow you run into the problem that the water doesn't spend enough time in the rads to properly cool itself off plus all the heat dump from the pump, So the graph beween temps vs flow is something like a U shape where there is an optimum flow for the best temp.

What you can do is have a regular pump running the rads to a common res then another super pump set up like the RD-30 or 3x DDC 3.2 from the same res to the cpu loop to cycle as much heat off the cpu as possible. Again this is another overkill method because no matter what you do you will never go below ambient and it may only net 2-3C at most better than a conventional set up.

i don't think is worth all the money dump into a WC system that nets you 2-3C that is overbuilt and very susceptible to failure due to the high flow and preasure you pumping through, well i would look into phase or a water chiller as alternatives.

thank you for the great suggestions. i guess to be clear, i don't want to phase change (not because of cost) for two reasons: irritating buzzing of refrigerator thing and i don't know that i can cool several components with a phase change (not to mention the fear of condensation...i know it's all math, but still)

i want to cool several computers. if i built this tomorrow, it'd cool 2 separate computers plus their gpus, ram, chipsets, etc. in the future as i add more machines to my business i'd like to cool those as well. perhaps an illustration on the levels of the 7 legged spider will explain better.



from the pump it'd go to a pvc pipe sending the water to a pvc valve system that would have several open/close gates to allow the addition of more machines over time. then from each gate to a computer system. from those back to a pvc valve gate system, one return pvc line to the cooler with antifreeze in it, through the three radiators (you can see the water getting colder as it passes through each rad) to a reservoir to the pump and starts all over again.

so here in lies the question.
1) with a setup like this, will the force of the pressure (from the pump being in excess of 500gph) hurt the waterblocks or not?
2) if not, does that mean i need to optimize the pressure so the water absorbs the heat from the respective heat sources to cool adequately?
3) will water at a higher velocity cool more efficiently? or is it the parabola optimization here?

Originally Posted by Fafeifa

Once you go too high in flow you run into the problem that the water doesn't spend enough time in the rads to properly cool itself off plus all the heat dump from the pump, So the graph beween temps vs flow is something like a U shape where there is an optimum flow for the best temp.

./facepalm


This is wrong, completely wrong. The water spends the same percentage of time in the radiator no matter what your flow is. At higher flows the water will not cool down as much as it moves through the radiator, however, the water will also not heat up as much as it goes through the waterblocks. It is mostly a wash. In general however radiators perform better as the flowrate increases. Don't take my word for it, go look at the test charts for the MCR-320 on swiftechs site. You see as the flow increases so does the potential cooling capacity.


As for your pump, 500gph isn't all that much. The D5 has a max flow of 317 gph. A garden pump you get at a home improvement store might have high max flow, but its head pressure (ability to push against restriction) will most likely be lacking. Therefore in a restrictive water cooling loop your 500gph pump might even perform worse than a high head pressure pump like the DDC or D5. This is one reason why the Eheim pumps aren't used as much anymore, they aren't high head pumps and therefore don't quite have the performance of the newer watercooling pumps. They are still very quiet and very reliable pumps though.


Might I ask how you plan on cooling your "tub o antifreeze"? You can't just stick some radiators in some antifreeze and think that will cool them. The antifreeze will heat up just like anything else will. You might want to consider a geothermal cooling loop.

EDIT: Was going to link to Naja002's thread, didn't know whose thread it was, then realized that you referred to his setup in your original post. Still confused what your green box o antifreeze is for.

cooling the tub of antifreeze is simple (the way they do on cars) two 320mm fans on top. either that or have the antifreeze on it's own loop but i think that is a silly idea. geo thermal unfortunately is not an option. it's a nice thought but i cannot do it.

Well if you just throw two fans on top blowing on the antifreeze it's not going to cool it enough for it to not end up worse than a normal liquid to air loop.

Like the other guy I really don't get what your are trying to do with the antifreeze?

Looks like your trying to cool your rads with it rather then using fans on them?

If so why?

Why add the extra step?

What do you hope to gain from transferring the heat from one fluid to another before transferring that heat to the air?

If the loop your trying to build to cool the rads can really remove the heat better then the one cooling your PC then that loop should be the one cooling the PC directly.

Originally Posted by Erasmus354

./facepalm


This is wrong, completely wrong. The water spends the same percentage of time in the radiator no matter what your flow is. At higher flows the water will not cool down as much as it moves through the radiator, however, the water will also not heat up as much as it goes through the waterblocks. It is mostly a wash. In general however radiators perform better as the flowrate increases. Don't take my word for it, go look at the test charts for the MCR-320 on swiftechs site. You see as the flow increases so does the potential cooling capacity.


As for your pump, 500gph isn't all that much. The D5 has a max flow of 317 gph. A garden pump you get at a home improvement store might have high max flow, but its head pressure (ability to push against restriction) will most likely be lacking. Therefore in a restrictive water cooling loop your 500gph pump might even perform worse than a high head pressure pump like the DDC or D5. This is one reason why the Eheim pumps aren't used as much anymore, they aren't high head pumps and therefore don't quite have the performance of the newer watercooling pumps. They are still very quiet and very reliable pumps though.


Might I ask how you plan on cooling your "tub o antifreeze"? You can't just stick some radiators in some antifreeze and think that will cool them. The antifreeze will heat up just like anything else will. You might want to consider a geothermal cooling loop.

EDIT: Was going to link to Naja002's thread, didn't know whose thread it was, then realized that you referred to his setup in your original post. Still confused what your green box o antifreeze is for.

Actually, it is not completely wrong. In order for water cooling to be MOST effective it actually needs to spend some time in the area it's trying to cool, obviously not the radiators. If the liquid is moving too quickly it will not be able to absorb the heat. Eventually this will lead to elevated temps.

This is the same principal your car works on. If you don't believe me, take out your thermostat and see how quickly it overheats.

when i saw naja02's post about the chilled cooling
http://www.hardforum.com/showthread.php?t=1424727
also his idea here as well
http://www.hardforum.com/showthread.php?t=1421685

essentially combine parts of both cooling ideas into one.

i thought, there's an idea but why go to the extreme of using an AC unit if i have the fans exhausting heat out of the rads rather than pushing air in, it'd probably be better to have something circulate the antifreeze as well. i have a relative grasp on the laws of thermodynamics (since the original ending to battle angel alita Thanks Desty Nova! )

i understand about getting rid of heat. i know the water has to be in the waterblock long enough to grab the heat and carry it away. been working on cars for years so i understand the heat exchange that needs to take place. my main goal is not the tub of antifreeze. my main goal is to have a strong enough pump to run one loop to several computers and their smaller components.

http://www.koolance.com/water-coolin...product_id=372

they have a system that can cool many computers per loop, if they can do it, then i should be able to work the math out and do the same thing.

i know that water rushing across the block "washing" i believe i read someone call it in this thread, is not going to accomplish a lot. so it'll be a lot of me having to adjust settings in my plumbing so i don't hit some sort of entropy.

Is this a bad idea?
should i spend more money on individual loops per computer system, why?

Originally Posted by Erasmus354

./facepalm


This is wrong, completely wrong. The water spends the same percentage of time in the radiator no matter what your flow is. At higher flows the water will not cool down as much as it moves through the radiator, however, the water will also not heat up as much as it goes through the waterblocks. It is mostly a wash. In general however radiators perform better as the flowrate increases. Don't take my word for it, go look at the test charts for the MCR-320 on swiftechs site. You see as the flow increases so does the potential cooling capacity.

So you are saying that flow does escalate infinitely?

Originally Posted by Fafeifa

So you are saying that flow does escalate infinitely?

IIRC there are more pumps operating as GPM increases on that graph, so what you're seeing is the heatdump from the extra pumps pushing the radiator used in that test over its dissipation limit.

Originally Posted by OmegaSupreme

This is the same principal your car works on. If you don't believe me, take out your thermostat and see how quickly it overheats.

I'm going to disagree with that part. I know for a fact we had major overheating issues in Afghanistan. Going up steep hills in armored hmmwv's at high altitudes was a major problem. One of the things we did to help fix it was remove the thermostats. It helped, didn't fix it but trying to run up hills like that with a few thousand pounds more than you should isn't something you can easily fix.

In regards to one of your original questions about it damaging the blocks etc. I think you're going to see damage occur at your weaker points first. The tubing and joints etc. If your tubing is holding up to it then I wouldn't worry about the blocks etc at all. I'm not a total expert on this type of thing but that seems the common sense answer to me.

I don't see a high rate of flow within reason taking away from the cooling properties either. The radiator / blocks are going to be constantly exposed to water, that water is going to be swapped between the two. I think that the cons to a high speed are close to nill. Obviously if we start talking about super high speed and friction and all that it would apply but that's pointless because there is no reason and your hoses would burst first.

I'm open for more debate, not claiming to be 100% but I think it shouldn't be a problem.

That test was using 2x MRC 320 that is plenty of radiator for a i7 and few pumps. But you are right the more flow also brings more heat dump. Although i doubt a few pumps will put any sweat to the rads.

Originally Posted by nomu

IIRC there are more pumps operating as GPM increases on that graph, so what you're seeing is the heatdump from the extra pumps pushing the radiator used in that test over its dissipation limit.

SO avoid too much pressure (i'm in agreement about the thermostat, my buick was overheating like crazy last year because it was stuck closed, removed the thermostat, forgot to replace it. ran fine ever since, it runs a little warmer in the summer, but warmer in the winter as well and that's not a bad thing per se). avoid ridiculous hose bends, probably should use the thread reinforced tubing. on my radiator submerged setup, i should have something circulating the coolant plus exhaust fans and something to probably to cool that as well. next, make sure that the garden pump has enough high end head room to force water along even if there is flow restriction. then after all of that, make sure my joints and clamps aren't bulging and busting.

sounds like everything is in order. i guess my original question has been answered. yes i can do it. the majority of you out there think it seems like more work than necessary. and be careful. remember to post results? something like that?

Originally Posted by StoneTZ

I'm going to disagree with that part. I know for a fact we had major overheating issues in Afghanistan. Going up steep hills in armored hmmwv's at high altitudes was a major problem. One of the things we did to help fix it was remove the thermostats. It helped, didn't fix it but trying to run up hills like that with a few thousand pounds more than you should isn't something you can easily fix.

Well you have to understand that as your altitude increases, the temperature at which water will boil also drops. The thermostats in your humvee's were probably rated @ 180*F @ sea level. All the sudden you're at 7500 feet and your vehicle is overheating. In this case removing the thermostat would be more beneficial as it would take considerably longer to heat up all of the coolant rather than what was in the block. And keeping it circulating would not give it the opportunity to absorb the heat. Therefore, keeping it from overheating. Although, given enough time this setup would eventually overheat as well. Ideally, they would've installed a lower temp thermostat and that would've remedied the overheating situation.

To clear up the debate:

High flow vs. low flow does not matter. The same amount of heat is transferred either way with the exception of crazy high flow attributing to friction. Liquid moving quickly through a loop would have roughly the same temperature all throughout. Liquid moving slowly would be hotter at the waterblock and cooler at the rad (larger delta T). In terms of heat dumped, it should come out the same either way.

It'd do better to get a higher PRESSURE pump and add more liquid/rads to the equation. Takes longer to heat up a larger amount of water, and extra rads would cool it off faster.

To be on topic, I assume the only reason the OP wants this complicated setup is to save money on radiators for multiple comps? You could probably use an automobile radiator for the "big tub o' antifreeze" with some sort of slow moving pump to circulate and cool the tub. Without anything cooling the tub it would take awhile but eventually the temps would not be very appealing. It's an interesting idea, similar to how air conditioners in giant buildings work, I really don't think it's practical to do unless you're cooling 4+ computers or you just want to prove the concept.