How To 

The influences of water flow rate in water cooling.

(All of the statements below assume that there’s a system at thermal equilibrium with constant load and constant ambient air temperature.)

Assuming the water cooling circuit consist of one convector(*), a pump (with reservoir) and one or more cooling blocks the flow rate has one major impact: The difference in water temperature between the warmest and coldest part of the loop.
The temperature difference (in degrees Kelvin or Celsius at equilibrium) is given by 0.86*P/F where P is the heating in watts and F is the flow in litres per hour. (Replace 0.86 with 1.55 if you want it in Farenheit.)

As can be seen, with a high flow the difference in water temperature is low and you don’t need to consider in what order to mount the components.
I've also seen mentioning that with higher flow you get more turbulence and thus better heat transfer and better cooling. I think this effect on the cooling is negligible. Cooling blocks are designed to create sufficient turbulence as is, and extra turbulence is actually energy added to the water that also needs to be removed in the convector.

With a low flow things change though:
* When the water cooling is used to cool only one part you can have a fairly low flow without reducing the cooling efficiency. If the pump then is mounted on the ”warm” side it might be subject to water that’s a bit warmer than what it’s rated for.
* With more parts cooled off, like CPU and GPU, the first cooling block after the convector will be fed cooler water than the second (and third) block. When the difference is in tens of degrees this will be quite notable in the component temperatures!
* Also notice that if you have a ”small” loop with lowish flow and then add more parts the flow will be further reduced while heating goes up, increasing the temperature difference a lot!

How does it effect the actual cooling?
* As long as the difference between ”cold” and ”warm” water in the loop isn’t much more than five degrees the difference between average water temperature and ambient air won’t differ measurably. Thus the cooling efficiency doesn’t change as long as the water flow is high enough. Increasing it further won’t help.
* When there’s a high variation in water temperature throughout the loop (with a low water flow) it’s elementary that the ”cool” water will stay a bit above ambient air. The convector efficiency is also somewhat reduced even at higher average temperatures. The impact of more than one cooling block is discussed above, with the second part recieving reduced cooling.

Discussion
Blanket statements like ”pump speed doesn’t matter” and ”tube diametre doesn’t matter” are obviously false, since these two factors influence the water flow. When a weak pump is used both decreasing the pump speed and opting for thinner tubes will reduce the water flow and thus increase the temperature difference within the loop.

The statement that ”the order of components doesn’t matter” is only true given that the water flow is high enough.

The obvious recommendation is to not have a very low flow. If your flow in litres per hour is the same as the combined heating in watts you have enough flow. If the water flow is considerably lower than that you need to consider the effects mentioned above.

To avoid low flow you can:
* Use thicker tubing.
* Use smooth curves rather than sharp angles in tubings and fittings.
* Use cooling blocks and convector that are less restrictive to the water flow.
* Use a strong pump, like for example a D5.
_________________________________

(*) I use the term ”convector” rather than the more common ”radiator” because the latter imply that (most of) the cooling is done by black body radiation rather than being influenced by air flow.
A convector use airflow to transfer heat.
 

Olle P

Distinguished
Apr 7, 2010
720
61
19,090
References
I was asked about references for the claims above, and here's my respond to that:

This article is based on fairly straightforward physical laws and correlations that could be seen as common knowledge, and a tiny bit of afterthought and reasoning combining those correlations.
Add a bit of basic math and you get the presented result.

First paragraph:
The only piece of firm data is the thermal capacitivity of water, which you can find in many sources.
Combine this with the laws of thermal dynamics and you get the provided result.

Second paragraph (high flow):
Pretty much self explanatory. Apply general laws of physics.

Third paragraph (low flow):
Also very self explanatory. Just apply logical thinking.

Fourth paragraph (actual effects):
  • First bullet: Derived from a bunch of water cooling reviews as well as from the second bullet and common sense.
  • Second bullet: "The convector efficiency is also somewhat reduced [when having a large delta-T for the water] even at higher average temperatures."
This is derived from an equation I found in a product sheet for regular radiators (for heating rooms) years ago.
The effective thermal "average" temperature of the water, used to calculate the heat given off to the room, isn't exactly half way between the inlet and outlet temperatures but closer to the outlet. This deviation from the arithmatic average is increased with the increase of temperature difference between inlet and outlet. With the relatively small change in water temperature normally present in a computer cooling setup it's pretty much a non-issue, but more relevant in a central heating system.
If you really need some form of reference or further proof of this look at data and fomulae for regular radiators to find evidence that a low flow is less efficient for giving off heat.

Conclusions:
Self explanatory, based on the previous info.

Recommendations:
First three bullets are all common knowledge to reduce the flow resistance.
____
Further comment
Notice that my article leave out the most important aspects of successful watercooling:
A properly mounted and well designed cooling block (to maintain a low temperature difference between CPU and water) combined with a sufficient convector (to maintain a sufficiently low temperature difference between the water and ambient air).