• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!


Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.


Mar 20, 2001
Just before I started watercooling I noticed a lot people always said that more water flow means better cooling.

But if water moves too quickly won't it not have enough time to cool?

I cut my hose lenghts in half to fit the setup into my case. Just to check what effect this would have I left everything outside so the only difference was the hose lenght. My temperature went up!

I'm guessing the smaller hose lenght reduced the time it needed to get from one part of the setup to another. Basically, the speed increased.

Reason for edit: Now that it's stickied I have to watch my spelling :D
Last edited:
This has been a debated subject for quite a while now. I personally think that with a high flow rate, a bigger radiator with better cooling is necessary.
Well, this is just what sparked up in my head while reading this post: If the water moves it will have less time to pick up heat. I think the key is finding the sweet spot (the proper flowrate for your setup) to save yourself that extra degree of water temperature.
Real World tests on other sites have shown that higher
flow rates are not always better. By shortening your hoses
you may have increased your flow rate. This might have
bumped you out of your sweet spot.

Note: we are talking about a "system" here. If you
isolate the WB or the radiator or any other part for
your analysis you may be lead to a false conclusion.

Generally, you want a big, low restriction radiator.
Well, being the newb that I am I didn't realize that my system hadn't been completely bled, and that's why my temperature rose.

However, I think that finding a "sweet spot" will be very difficult since most pumps don't come with a flow rate control nob.:confused:
I dont recomend this, however, you could partialy kink your hose to slow the water flow. If you really happen to be a tweaker you may wish to buy a flow regulator to make those manual adjustments.
There is an elementary equation from basic thermodynamics that states that the rate of heat transfer (Q) equals the mass flow rate (M) times a constant (the specific heat of water) times the delta T (fluid temp out minus fluid temp in).

Q=M x c x Delta T

In other words, the rate of heat transfer is directly proportional to mass flow rate. You increase the flow rate, you will then increase the rate of heat transfer. Since you cannot mess with mother nature it is very naive to think it works any other way.

Assume the CPU inserts a constant rate of energy (Q) into the cooling system. Then, from the relationship above, increasing the mass flow rate must result in a smaller delta T because Q remains constant. This smaller Delta T (fluid out - fluid in) also means that the average fluid temperature in the water block is somewhat lower even though the rate of heat transfer has not changed.

Now lets look at the heat transfer from the CPU to the water. The rate of heat transfer between two points is proportional to the temperature difference between those points. In our case this Delta T (not to be confused with the one above) is the temperature of the CPU minus the average water temperature in the water block. Lowering the average water temperature, as we did above by increasing the flow rate, means we have a little better heat transfer from the CPU to the now somewhat cooler water. The result is that the CPU runs a little cooler.

This all says that if you increase the flow rate, and everything else remains constant, you will decrease the CPU temperature. However, everything else will not remain constant if you increase the flow rate by using a larger pump. The pump uses some amount of electrical energy. This energy must end up somewhere. A relatively small amount of it is dissapated as heat from the motor. The overwhelming majority of it is converted from electrical energy to mechanical energy in the form of a rotating shaft that does real work on the water. This energy ends up in the water by increasing its temperature. It is called "pump heat" and can be very significant. An Eheim 1048 is rated at 10 watts, almost all of which ends up in the water. I understand a very overclocked CPU is good for upwards of 75 watts. As you can see a smaller pump like the 1048 contributes about 13% to the total heat load on a system with an energy hungry CPU. With other more common CPUs running at 25 to 50 watts, this percentage is much higher and is therfore much more significant.

As an interesting aside for those non-believers, this is also why excessive use of a blender to mix up frozen orange juice results in the juice not being as cold as expected. Also, nuclear power plants use primarily pump heat (from three or four 6,000 HP pumps) to heat up almost 75,000 gallons of water from 200 degrees F to about 550 degrees in about six hours or less.

The point here is that there is a trade off in how big a pump to use to increase the flow rate. More flow is beneficial. It is best to achieve the desired flow with a small a pump as possible and flow paths with minimum flow resistance. The bigger the pump, the more heat is added to the system. Eheim makes a 50 watts unit that I see talked about every now and then. This guy is probably a bigger heat load on the cooling system than the CPU itself.

Bottom Line: If you increase flow rate with the same pump your temperatures will trend in the direction of goodness. If you increase flow rate by going to a bigger pump you will reach a trade off somewhwere where the pump starts putting too much energy into the system and temperatures will start increasing.

I did not intend this to be so long but I do hope this helps remove some of the confusion from this issue.
You don't want it to flow too fast because you want the water to cool inside the radiator.

But then again you don't want too little flow because the water gets warm inside the waterblock and if it doesn't flow fast enough you won't get the best cooling you should be getting.

It's really what you think is right, after-all it is your watercooling system. ;)
kttdkt, trust me, more flow, more cooling.

If you slow the flow rate rate so the water sticks around longer it will indeed absorb more energy and increase to a higher temperature before it leaves the water block. HOWEVER, the warmer the water. relative to the CPU, the slower the heat transfer rate. Put another way, each hunk of water will leave the water block with more energy but at the expense of a lower heat transfer rate.

By increasing the flow rate, the water does not absorb as much energy but its average temperature in the water block is lower. The lower temperature means heat transfer is happening at a greater rate. The higher flow rate quickly moves water out of the block and allows new cold water to enter, maintaining a colder average fluid temperature and a higher rate of heat transfer.

This concept also applies to the radiator. It is a heat exchanger just like the water block. It only has a different geometric form.

Unfortunately, most people are let down by their intuition when it comes to heat exchangers.
Last edited:
The forum ate my post, and didn't spit it back out onto the web!! I hate it when that happens.

Anyways, the gist of my post was :

How much heat does an Eheim 1250 317 gph pump really add compared to a Eheim 1048 148 (?) gph pump? I doubt it's much more.

RhoXS, wouldn't having a slower pump allow more time in the radiator also thus allowing more time for heat exchange there?
neo86, the Eheim 1250 is rated at 28 watts according to Eheim's web site (www.eheim.com). The 1048 is rated at 10 Watts. I have to assume this rating is at maximum flow or "runout" as Eheim does not specify at what flow rate this rating applies. A centrifigal pump uses more energy when it pumps more water. In practice the actual flow is not at maximum because of the flow resistance etc. in the flow path. As a result, the actual heat input from these pumps is probably somewhat less than the specified rating.

Q=mcDT applies to all heat exchangers (as long as there is no phase change - e.g. steam to water etc.). The highest rate of heat transfer across any substance occurs when the temperature difference (Delta T) is greatest. As the fluid is cooled in the radiator its temperature starts to approach that of the cooling air. As the fluid cools, the rate of heat transfer decreases because of the decreasing temperature difference. Since the sole purpose of the radiator in life is to remove heat from the system, the faster it transfers heat the better. I have not taken any measurements to prove this but it appears to me in my system that the discharge temperature of the water approaches that of the cooling fluid (air). This means that with any reasonable flow the difference in temperature at the discharge of the radiator between the fluid and air is relatively small. This also means that there is already little heat transfer taking place at this point. Decreasing the temperature incrementally more (by "slowing it down") will not result in any consequential increase in the total amount of energy removed. However, since less mass moves through the radiator at a slower flow rate, less total energy is removed.

The mathmatics associated with basic thermodynamics is a gravity issue. Heat transfer rate is directly proportional to flow rate. Ones intuition can be misleading here because it is very easy to neglect certain key principals. The math, properly applied of course, will always get you the right answer.

I have seen real world exanples of this principal and it really is true. A system I am familiar with, as an example, uses two 600 HP parallel pumps (one normally in standby) for pumping water from a number of heat sources to one very large heat exchanger. The large heat exchanger uses ocean water for cooling. On hot summer days the heat load on the system increases somewhat and the temperature of the ocean water starts approaching that of the fluid to be cooled. The second pump (normally not running)is run to increase the flow rate. The flow rate does not double but it does increase. System temperatures do indeed drop even though the fluid is moving through the heat exchanger at a faster rate. The reason, as above, is because a higher delta T is being maintained between the cooling fluid and the fluid to be cooled. Also, more mass flow through through the heat exchanger means that there is more water to give up energy to the cooling fluid at this higher Delta T.

I think I beat this to death so this will be my last attempt to explain this. I hope this helps.
neo86 said:
RhoXS, wouldn't having a slower pump allow more time in the radiator also thus allowing more time for heat exchange there?

Neo, you do NOT want the water in the radiator very long. The longer the water is in the radiator, the cooler it gets. That's good, right? Wrong. It decrease delta T between the water and the cooling air. Decreases in delta T mean efficiency will decrease. If you want your radiator to dissipate the maximum amount of heat, let the water get real hot. That's good, right? Wrong. That will decrease the delta T between your CPU and the water, decreasing efficiency there. If you want to dissipate the most amount of heat at your CPU, get the water as cold as possible. You want to do this again? Cause it goes in circles. The "sweet spot" is a natural equilibrium that will occur based on CPU temp, ambient temp, and water and intake air flow rate. RhoXS is 100% correct, more flow is better, hands down, no argument. Physics does not lie. For both water and air.

He is also right on with the heat from the pump, cruel world, isn't it? More flow, more better, more flow, more heat. Find the balance. The beautiful thing is that it takes a lot of energy to heat up water. As the water flows past the CPU, it does not pick up that much energy, so temp does not increase that much per pass. This means that small flow rates are fine on the CPU side, and increasing rates will decrease returns quickly. Radiators are different though. There is a lot of surface and tube to flow through in a rad, so temp will drop significantly in a rad per pass. So high flow is more important for rads.

One thing also that I should mention, intake air is the same concept. More flow, the better. And since the air is exhausted and not recycled(I'm not gonna get into your room as a system, in which it is recycled, so we can assume constant ambient temp) "pump heat" for air is negligible. More air es bueno. Hands down, no argument. If you can increase air flow or rad surface area to the point that it will drop rad temps to near ambient, that is very good, but very hard. As rad temps decrease, delta T decreases, decreasing efficiency. Thermodynamics sucks, huh?

The lesson from this? Increasing flow rate is good, for both air and water. Air is better since it has less consequence for increases in flow, no "pump heat". But both increase rad efficiency. Hope this helps, RhoXS did a pretty good job, but I was bored and felt like kicking in support for him.
flowrate is better

unless it changes boundary conditions for the worse....

I'll go back into my cave now...
Heh, I'm glad I'm wrong because I'm using an Eheim 1250 pump :D

I'm just glad someone finally started in with the physics of this to clear it up. You laid a really good foundation, I just came in to clarify. There have been many things said about flowrate, and I disagreed with most of them, but did not want to get into it. Those debates can get pretty "heated" and go nowhere. Not enough flow rate :D You're right about the counter-intuitiveness of this stuff.
You're right about the counter-intuitiveness of this stuff.

That is exactly why I felt compelled to try to explain it. Unfortunately, this thread will quickly dissappear from the first few pages of this topic and the same issues will again quickly arise.

One last thought. There seems to be a very strong tendency among everyone interested in this subject (including myself) to achieve as low a CPU temperature as possible. This is really not good engineering as the goal should be as low a temperature as necessary for stable and reliable operation. Anything beyond that serves no real purpose. However, I have to readily admit that there is definetly satisfaction in making a mod and seeing it result in an even lower CPU temp.
This might almost deserve a sticky, if the mods feel that it explains the mystery of flowrate well enough. It's one of those things that everyone should know, but very few do. There is a lot of misinformation out there too. And lower temp is better, necessary or not :) It makes components last longer, and if you get it colder than you need to OC to a certain level, push it to the next. One more clock tick per sec would justify it to me.
I want to bump this back to the top of the board and put in my vote for it to be made a sticky. It answered some questions for me very well.

Major Bump !!!! Anyone who doesn't read this is losing out.
Not to hijack your thread Neo but I have a question after reading this. If you aren't familiar with the thread I have going it's The Twister bong. Now I am guessing when you were talking about the pump itself generating heat and adding it's energy to the coolant did you mean all pumps or just submersable types. I've fought with this for a long time possibly because I'm stupid. my system uses one or two pumps 1600 to 1800 GPH (new style washing machine pumps). It has worked so well at cooling water it baffles me. I have a high flow rate of the water as well as air. My system sometimes will all of a sudden seem to gain great amounts of heat I tweak the flow rate and it starts to slowly respond............. I think possibly I might have a clue as to why. I got to read this again too much information for one read . Great thread!!!!!!!!
Stay Cool