The Myth of Turbulence in Water Cooling

aka Bob is assimilated by his Borg Cube...

Anyone who has done much reading on the Internet about water cooling will come across discussions of turbulence before too long. Arguments are made that the turbulence will increase flow rates, result in improved heat transfer, and a whole host of other benefits. While there may be some truth to some of this, it is far from proven. From my perspective, this turbulence argument is advanced far too often simply as an argument for why a home made block made by drilling consecutive holes to create the water passage is better than a CNC milled block. To put it simply, I've got my doubts

 

Frequently the argument is made that dimples are good for increasing flow because they work so well on golf balls. It is true that a golf ball's dimples will just about double its efficiency from a combination of reduced drag and increased lift. But the reason this works for golf balls has very little to do with water blocks, it seems ot me. Let's consider what happens. Air flowing over any surface that's in motion (either the air or the object is in motion) has to make a transition from the relatively stationary surface to the full speed. For a golf ball, air right at the surface is practically not moving at all, while if you move a few paper thicknesses out from the surface its moving very rapidly indeed. A well hit ball can fly through the air at 200 km/hour! So how do the dimples work.

For purposes of this discussion, assume the ball is stationary, and 200 km/h air is blown over it. A perfectly smooth ball actually "grips" the air a little better, thickening the layer of slower moving air. This makes the teardrop-shaped boundary layer of slow moving air relatively large. The air has to move out quite a ways before its moving at full speed. This larger teardrop has more surface area, and hence more drag. There is also a partial vacuum behind the ball that tends to drag it away from the direction of travel, slowing it down further. Adding the dimples creates turbulence, which breaks the air loose from the surface of the ball and gets it moving. The resultant teardrop-shaped boundary area is smaller, and drag is lessened.

What about this principle being applied to things other than golf balls? Well, there are a number of differences here. First, the ball is spherical, which is actually not a very aerodynamic shape. Teardrops work best, and you can imagine why looking at the shape of the boundary layer. Also, the ball is spinning to create lift and this also affects the aerodynamics. Most successful use of turbulence is aimed at cases where the shape of the object was not accomdating a smooth boundary layer transition. Consider an airplane wing. Taking a ball peen hammer to it to increase turbulence over the surface is not beneficial! However, there are a number of turbulence-inducing aerodynamic aids for wings, such as wing fences and vortex generators. Their goal is to use turbulence to manage airflow once the wing is operating at a speed or angle of attack where the boundary layer starts creating problems (usually stalling the wing). They're great for raising stall speeds, but seldom result in less drag, higher speeds, or better fuel economy.

What about water blocks? Swiftech, for example, has a dimple pattern in their block, and it performs pretty well:

In this case, it isn't surprising the dimples help. For one thing, they increase the surface area. For another, it probably is a good idea to get some turbulence going in this thing or that water will just blow through without picking up much heat. On the other hand, I really doubt those dimples increase flow rates. Many water blocks are more similar to the maze and spiral designs where the water follows a path that has been cast or milled into the copper. As I mention above, many are homemade blocks where the enthusiast did not have access to a milling machine, and simply drilled a line of holes. Sometimes both a mill and drilled holes are employed. Here is a gorgeous example of one such block:

Is this design any better than a smoothly milled path through the block? It's hard to say owing to the lack of back to back testing where other variables are controlled, but I doubt it. The Gemini Cool is just a plain milled path and tests extremely well in head to head reviews:

It's kind of a shame nobody has bought 2 identical blocks and just tested out this turbulence = better flow rate + better cooling hypothesis. It would be the easiest thing in the world to do, and not all that expensive. Buy a pair of Maze 3 blocks at $40 apiece. Keep one as the "control" in the experiment and dimple the heck out of the other. Compare the flow rates and temperatures to see if one is helping or hurting.

Back to Water Cooling Central...

 
All material 2001-2006, Robert W. Warfield.