This material has been heavily edited. Any resultant errors are solely my fault. Ed
From Mike Larsen:
You recently published and article written by Allen Stace about water
cooling and his views on it. He said:
“The most efficient choice is more surface area at a slow flow rate.
That’s right, SLOW. You have to give the water enough time to dump all that
heat and a lot of cool surface with which to do it.”
Yes, with a slower flow rate, water will be able to dump more of its heat into
the walls of the radiator. However, because of the slower flow rate, less
water will get the chance to dump its heat.
I have been working on designing my water cooling system so let’s take a
look at what I’ve been doing.
I want to design a system that provides the most cooling in the least amount of space
dissipating as much heat as possible efficiently. To do so, I’ve focussed on maximizing the length
of space where the water can be cooled.
Here is the basic equation for heat transfer through a tube
(It doesn’t consider the conductivity of the material of the pipe, the efficiency
of fins attached to the pipe, volume of air flowing over radiator, etc.):
qdot is the heat transfered in watts per meter cubed
Dout is the outside diameter of the tube
Din is the inside diamter of the tube
L is the length of the tube
mdot is the mass flow rate of the water in kg per second
cp is the specific heat of water at a given temperature
T2 is the temperature of the water leaving the tube
T1 is the temperature of the water entering the tube.
(Excuse the mess, tough to type equations in just a text capable e-mail message!)
What you need to understand is that water flow AND temperature difference are on the
same side of the equation. If you decrease one to increase the other, you are working
against yourself. It’s like changing 4X2 to 8X1; you get the same result.
However, this equation isn’t necessarily just like a choice between 4X1 and 2X2. If you found out
that you if you tripled the water flow, and got half the original heat reduction; it would be like
changing 4X2 to 12X1. You’re better off (though not three times better off).
On the other hand, if you halved the water flow, and found that the difference in temperature increased
by 50%, it would be like turning 4X2 into 2X3. You’d be worse off.
I personally believe that at least to a point, more waterflow will more than offset any reduction in temperature difference.
There is a definite point of diminishing returns, though. A while back,
a link was posted to a site that looked at temperature differences in the CPU
as water flow increased. With every increase in water flow, there was a decrease in CPU
temperature. It was also very easy to see that the first increase of 50 gph
had a much bigger impact on temperature than the next 50 gph increase.
Therefore, there is a ‘sweet spot’ of how much gph you can push through to
get good cooling, vs. much higher costs of equipment and higher risk of failure from
leaks and other malfunctions.
Something else to consider is that water has to move fairly quickly through the tube
to get turbulent (better for cooling) rather than laminar
water flow. What does this mean to the average Joe and
his water cooling setup? Since every system is different, and there won’t be any
government studies soon, you will just have to take your system and experiment with it.
In no way at all am I trying to downgrade Mr. Space’s
article as it has some great information in it, but I thought that this
particular point might need a little clearing up. I’ll continue my work
and maybe even write a paper on the results. There are many other factors
that affect cooling (effective surface area of water block and radiator, ambient air temp/humidity/changes, etc…) My water block is currently still in the design phase but should be
headed to the machine shop fairly soon. When I get it, I’ll take some
measurements on temperature differences vs. flow rate.
Until then, happy
M.S. Mechanical Engineering
Some comments from Allen Stace:
My main point about slow flow rate was conserning the radiator alone and not the system as a whole.
When the pump is rated at, say 60 GPH like mine, the real flow rate will be governed by the most restrictive point in the system. In most cases, that is likely to be the water block.
From there on, the pressure of the water will drop do to less flow resistence.
Though the flow rate throughout the entire system is the same, water will run faster or slower in different parts of the system, depending on the space it has. You don’t need to lower the total volume in order to slow the water down. As you’ll see, what I did was to slow the water down by giving it more space.
While you could have a water block with bigger holes, that would likely give you less efficient laminar waterflow.
If you wish to use as little space as possible, I have gotten the best results by using a car heater core.
The one I am using is only 1.75 x 6 x 8in. It splits the water flow up into (6 first pass, 7 second) 0.1 x 1.2 in. serpentine channels using a flood box design similiar to multipass heat exchangers.
This lets water contact a lot of surface area without a long run of pipe or back pressure. You get around 250 sq. in. of surface. If you used 1/2″ ID tube instead, it would require over 13 ft to get the same.
The material is also much thinner than copper tube, so heat transfers more quickly. It would be fairly simple to cut one down if you wanted a smaller design. Just a propane torch and a litle patience.